Skip to main content

Full text of "Year book - Carnegie Institution of Washington"

See other formats

/;v ; : : ,. ,.;;.■■ 


Digitized by the Internet Archive 

in 2012 with funding from 

LYRASIS Members and Sloan Foundation 



Year Book 63 


Library of Congress Catalog Card Number 3-16716 
Garamond/Pridemark Press, Baltimore, Maryland 



Officers and Staff v 

Report of the President 1 

Reports of Departments and Special Studies 1 

Mount Wilson and Palomar Observatories 3 

Geophysical Laboratory 55 

Department of Terrestrial Magnetism 291 

Committee on Image Tubes for Telescopes 403 

Department of Plant Biology 411 

Department of Embryology 491 

Genetics Research Unit 575 

Cytogenetics Laboratory 603 

Bibliography 615 

Administrative Reports 617 

Report of the Executive Committee 619 

Report of Auditors 621 

Abstract of Minutes of the Sixty-Sixth Meeting of the 

Board of Trustees 639 

Articles of Incorporation 641 

By-Laws of the Institution 645 

Index 651 

President and Trustees 


Caryl P. Haskins 


Barklie McKee Henry 

Henry S. Morgan 
Y ice-Chairman 

Garrison Norton 

Amory H. Bradford 
Omar N. Bradley 
Vannevar Bush 
Walter S. Gifford 
Carl J. Gilbert 
Crawford H. Greenewalt 
Caryl P. Haskins 
Barklie McKee Henry 
Alfred L. Loomis 
Robert A. Lovett 
Keith S. McHugh 
Margaret Carnegie Miller 
Henry S. Morgan 
Seeley G. Mudd 
William I. Myers 
Garrison Norton 
Richard S. Perkins 
Elihu Root, Jr. 
William W. Rubey 
Frank Stanton 
Charles P. Taft 
Juan T. Trippe 
James N. White 
Robert E. Wilson 1 

1 Died September 1, 1964. 

Trustees continued 


Keith S. McHugh, Chairman 
Alfred L. Loomis 
Juan T. Trippe 


Henry S. Morgan, Chairman 
Amory H. Bradford 
Walter S. Gifford 
Carl J. Gilbert 
Crawford H. Greenewalt 
Caryl P. Haskins 
Barklie McKee Henry 
Robert A. Lovett 
Garrison Norton 
Richard S. Perkins 
James N. White 
Robert E. Wilson 


Omar N. Bradley, 
Henry S. Morgan 
Garrison Norton 
James N. White 




James N. White, Chairman 
Walter S. Gifford 
Alfred L. Loomis 
Henry S. Morgan 
Richard S. Perkins 
Elihu Root, Jr. 

Seeley G. Mudd, Chairman 
Amory H. Bradford 
Crawford H. Greenewalt 
Elihu Root, Jr. 


Alfred L. Loomis, Chairman 
Margaret Carnegie Miller 
William I. Myers 
Charles P. Taft 


Keith S. McHugh, Chairman 
Barklie McKee Henry 
Garrison Norton 
Charles P. Taft 


Juan T. Trippe, Chairman 
Barklie McKee Henry 
Richard S. Perkins 
Robert E. Wilson 


Former Presidents and Trustees 


Daniel Coit Gilman, 1902-1904 Robert Simpson Woodward, 1904-1920 

John Campbell Merriam, President 1921-1988; President Emeritus 1939-1945 
Vannevar Bush, 1939-1955 


Alexander Agassiz 


Seth Low 


George J. Baldwin 


Wayne MacVeagh 


Thomas Barbour 


Andrew W. Mellon 


James F. Bell 


Roswell Miller 


John S. Billings 


Darius O. Mills 


Robert Woods Bliss 


S. Weir Mitchell 


Lindsay Bradford 


Andrew J. Montague 


Robert S. Brookings 


William W. Morrow 


John L. Cadwalader 


William Church Osborn 


William W. Campbell 


James Parmelee 


John J. Carty 


Wm. Barclay Parsons 


Whitefoord R. Cole 


Stewart Paton 


Frederic A. Delano 


George W. Pepper 


Cleveland H. Dodge 


John J. Pershing 


William E. Dodge 


Henning W. Prentis, Jr. 


Charles P. Fenner 


Henry S. Pritchett 


Homer L. Ferguson 


Gordon S. Rentschler 


Simon Flexner 


David Rockefeller 


W. Cameron Forbes 


Elihu Root 


James Forrestal 


Julius Rosenwald 


William N. Frew 


Martin A. Ryerson 


Lyman J. Gage 


Henry R. Shepley 


Cass Gilbert 


Theobald Smith 


Frederick H. Gillett 


John C. Spooner 


Daniel C. Gilman 


William Benson Storey 


John Hay 


Richard P. Strong 


Myron T. Herrick 


William H. Taft 


Abram S. Hewitt 


William S. Thayer 


Henry L. Higginson 


James W. Wadsworth 


Ethan A. Hitchcock 


Charles D. Walcott 


Henry Hitchcock 


Frederic C. Walcott 


Herbert Hoover 


Henry P. Walcott 


William Wirt Howe 


Lewis H. Weed 


Charles L. Hutchinson 


William H. Welch 


Walter A. Jessup 


Andrew D. White 


Frank B. Jewett 


Edward D. White 


Samuel P. Langley 


Henry White 


Ernest 0. Lawrence 


George W. Wickersham 


Charles A. Lindbergh 


Robert S. Woodward 


William Lindsay 


Carroll D. Wright 


Henry Cabot Lodge 


Under the original charter, from the date of organization until April 28, 1904, the following were 
ex officio members of the Board of Trustees : the President of the United States, the President of the 
Senate, the Speaker of the House of Representatives, the Secretary of the Smithsonian Institution, 
and the President of the National Academy of Sciences. 



818 Santa Barbara Street 
Pasadena, California, 91106 

Ira S. Bowen, Director 
Horace W. Babcock, 

Associate Director 
Halton A. Arp 
William A. Baum 
Edwin W. Dennison 
Armin J. Deutsch 
Olin J. Eggen 
Jesse L. Greenstein 
Robert F. Howard 
Robert P. Kraft 
Robert B. Leighton 
Guido Munch 
J. Beverley Oke 
Allan R. Sandage 
Maarten Schmidt 
Olin C. Wilson 
Fritz Zwicky 


2801 Upton Street, N. W. 
Washington, D.C., 20008 

Philip H. Abelson, Director 
Francis R. Boyd, Jr. 
Charles W. Burnham 
Felix Chayes 
Gordon L. Davis 
Gabrielle Donnay 
Joseph L. England 
Hugh J. Greenwood 1 
P. Edgar Hare 2 
Thomas C. Hoering 
Gunnar Kullerud 
Donald H. Lindsley 
J. Frank Schairer 
George R. Tilton 
Hatten S. Yoder, Jr. 


6241 Broad Branch Road, N. W. 
Washington, D. C, 20015 

Merle A. Tuve, Director 

Ellis T. Bolton, Associate Director 3 

L. Thomas Aldrich 

Roy J. Britten 

Louis Brown 4 

Bernard F. Burke 

Dean B. Cowie 

Scott E. Forbush 

W. Kent Ford, Jr. 

Stanley R. Hart 

Olaf Hartmann 5 

Cinna Lomnitz 

Brian J. McCarthy 6 

Alois Th. Purgathofer 7 

Richard B. Roberts 

I. Selwyn Sacks 8 

Ulrich Schmucker 

T. Jefferson Smith 

John S. Steinhart 

1 Resigned August 31, 1963. 

2 Appointed September 1, 1963. 

3 From May 15, 1964. 

4 From July 1, 1963. 

6 From May 16, 1964. 

6 Resigned April 30, 1964. 

7 From March 16, 1964. 

8 From February 16, 1964. 


Staff continued 


Stanford, California, 94305 

C. Stacy French, Director 
Jeanette S. Brown 
David C. Fork 
William M. Hiesey 
Harold W. Milner 
Malcolm A. Nobs 


115 West University Parkway 
Baltimore, Maryland, 21210 

James D. Ebert, Director 
David W. Bishop 
Bent G. Boving 
Donald D. Brown 
Robert L. DeHaan 
Irwin R. Konigsberg 
Elizabeth M. Ramsey 
Mary E. Rawles 


Cold Spring Harbor 
Long Island, New York, 11724 

Alfred D. Hershey, Director 
Elizabeth Burgi 
Barbara McClintock 

Cytogenetics Laboratory 
Ann Arbor, Michigan 
Helen Gay 

Staff continued 


1580 P Street, N. W., Washington, D. C, 20005 

Caryl P. Haskins President 

Edward A. Ackerman Executive Officer 

Marjorie H. Walburn Assistant to the President 

Ailene J. Bauer Director of Publications 

Lucile B. Stryker Editor 

James W. Boise Bursar; Secretary-Treasurer Retirement Trust 

Kenneth R. Henard Assistant Bursar; Assistant Treasurer Retirement Trust 

Donald J. Patton Administrative Associate 

Richard F. F. Nichols Executive Secretary to the Finance Committee 

Marshall Hornblower Counsel 

Staff Members in Special Subject Areas 

Tatiana Proskouriakoff 
Anna 0. Shepard 

Staff continued 


Carnegie Research Associates 

Paul W. Gast 

University of Minnesota 

J. D. McGee 

Imperial College of Science and Technology, University of London 

Research Associates of the Carnegie Institution 

Louis B. Flexner 

University of Pennsylvania 

Harry E. D. Pollock 

Carnegie Institution 

Paul Ramdohr 

Heidelberg University (Research Associate of the Geophysical Laboratory) 

C. E. Tilley 

Cambridge University 

Report of 
tee President 

It seems to be the accepted doctrine that no man is indispensable. Perhaps this is 
true if the objective is . . . mere survival; but, clearly, the doctrine has no basis in 
truth if the objective is something more. . . . All of the great breakthroughs, to what 
we call progress, have been made by men who were, indeed, indispensable. 

Henry Allen Moe — Seed Money 

There is no such thing as freedom in the abstract, divorced from the realities of a 
particular time and place. Whatever else the conception may imply, it involves a 
power of choice between alternatives, a choice which is real, not merely nominal, 
between alternatives which exist in fact, not only on paper. Because a man is most 
a man when he thinks, wills and acts, freedom deserves the sublime things which 
poets have said about it; but, as part of the prose of everyday life, it is quite practical 
and realistic. It means the ability to do, or to refrain from doing, definite things, 
at a definite moment, in definite circumstances, or it means nothing at all. 

R. H. Tawney — Equality 

It is the first step in sociological wisdom, to recognize that the major advances in 
civilization are processes which all but wreck the societies in which they occur: — 
like unto an arrow in the hand of a child. The art of free society consists first in 
the maintenance of the symbolic code; and secondly in fearlessness of revision, to 
secure that the code serves those purposes which satisfy an enlightened reason. 
Those societies which cannot combine reverence to their symbols with freedom of 
revision, must ultimately decay either from anarchy, or from the slow atrophy of a 
life stifled by useless shadows. 

Alfred North Whitehead — Symbolism 

of radio energy was discovered in the heavens, in observations made 
with the twin ninety-foot dishes of the radio astronomy observatory of the 
California Institute of Technology at Owens Valley. When the object had 
been brought within the optical field of the great Hale telescope on Palomar 
Mountain, the spectrum of its light exhibited a redshift so great as to mark 
it a new outpost of our reach into the universe. So distant is it that exact 
specification of its remoteness is not yet possible, because the conditions of 
space itself are still insufficiently understood. But it is evident that the 
light that reached the Hale telescope this past year to bring the image of 
the new object had left its source when our universe was hardly half its 
present age. This immense penetration into space that the year has brought, 
which is thought to reach over a large fraction of the entire radius of the 
universe itself, climaxes a long train of triumphs in astronomy which within 
little more than twenty years has increased our view into space by easily a 
factor of ten, changing distances thought of less than three decades ago in 


terms of hundreds of millions of light years to reaches now reckoned in the 
billions. Even a decade ago, we could not have remotely imagined what we 
know today. 

During this same year, a new and striking method has been perfected in 
analytical genetics at the molecular level that has opened the widest and 
most exciting potentials for the further extension of our understanding 
there. It makes possible the direct comparison among different organisms 
of the actual ordering of the nucleotide bases on the chromosomes — the 
ordering that constitutes the "language" in which the information of 
heredity is coded and stored from generation to generation in all plants and 
animals and in man himself. The method allows the comparative examina- 
tion of genetic codes among organisms far too distantly separated biologi- 
cally to permit the use of ordinary genetic methods. It suggests extremely 
broad and novel ideas about evolution at the level of the genetic code itself. 
It adds a powerful new experimental tool to aid in deciphering the very 
processes by which the hereditary information stored in the chromosome 
can determine the constitution — biochemical, physical, nervous — of the 
developing body whose nature and shape it directs. It bids fair to shed new 
light on the structure and function of microorganisms vital to man's welfare, 
including some viruses that appear to be concerned with cancer. The new 
concepts that this advance suggests, and indeed the very ones upon which 
it is based, would have been well-nigh unimaginable two decades ago. 

In 1963, nearly two and one-half million specialists were at work in the 
United States. They comprised almost four per cent of all those employed 
in the nation. Included among them were a million and a quarter or more 
scientists, technicians, and engineers. Nearly three in every five were 
engaged in projects supported or sponsored by the federal government, or 
located within the government itself. 

During 1963 the Swedish nation established a science advisory council, 
chaired by the prime minister and including five members of the cabinet. 
Thereby it joined the United States, the United Kingdom, France, and 
Canada and Australia, India and Pakistan and Turkey and Thailand, to 
name but a representative sample of nations where official actions taken 
within the last few years have reflected a striking and important develop- 
ment of our time. On both national and international scenes, over the years 
since World War II, science in its relation to affairs has become significantly 
transformed from the instrument of public policy which it has long been to 
a new and subtly but significantly different position — that of an object of 
public policy. The consequences of this shift could be considerable in the 
years to come. The shift itself is diagnostic of changes in the relationships 
of science to the societies which support it and of which it has become an 
integral part that may run deep. 

It would be hard to assemble four more divergent arenas of thought and 
action than are represented in these four great developments, occurring over 


the span of little more than a single year. They hardly seem linked at all, 
save in their common emphasis upon vast and wind-swift growth in the 
substantive range and power and penetration of the science of our day, on 
the one hand, and in its sheer massiveness and complexity and the extent 
and the degree, as well as the kind, of its social impact, on the other. Yet 
joining these diversities, and underlying them, there may be at least one 
set of common themes that deserves a close reexamination in the context 
of our day. They link three apparently dissimilar elements that yet may be 
closely related — the processes of planning both in and for science, the 
complex of contingent choices by which that planning is brought about in 
practice and through which both the shape and the functions of science are 
constantly being mediated and determined, and, finally, the role of the 
qualified and prepared individual who in all the aspects of science itself and 
of activities affected by science must make such choices. 

There is a related point that also challenges examination. When the early 
scientific associations came into being in Europe, the first Industrial 
Revolution had not yet begun, and their initial relations were with an 
essentially preindustrial society. When they were first founded in America, 
it was a rural and primarily agricultural society, not an urban and industrial 
one, that they early served and where they matured — a society which at 
the national level called upon the aid of science primarily in relation to 
agronomic matters. 

The original social environs of science have changed profoundly in the 
years between, in this country and abroad. As Geoffrey Vickers has pene- 
tratingly remarked: 

Rapid industrialization, as we see it today, is certainly not a cross section in a linear 
progress of indefinite duration. On the contrary, it is probably a rapidly passing phase, 
in which signs of self limitation can already be clearly seen. Yet ephemeral though it 
may be, it is of critical importance in two ways. For first, it involves the irrevocable 
commitment of resources, physical and human, in ways which may determine the 
pattern of life for many generations to come. And secondly, during its dominance, 
insights won over the last two thousand years, virtually the whole of our accumulated 
knowledge of value, will be either confirmed in a form profoundly changed or lost, for 
some later age painfully to discover. 

What have been the effects both on and for science at the national and 
international levels of its activity, and how are they related to the subtle 
shift of science in its public context from instrument to object of policy? 
The influence of science and technology upon our nation is as old as the 
history of the nation itself; a good case could be made, indeed, that it is 
considerably older. An important part of that influence, of course, has been 
and continues as never before to be exercised through the products of 
science and technology. But it is important to remember that it is a very 
different aspect of the scientific way which throughout our history may have 
exercised the deepest and most important influence on that profoundest of 
all our national concerns — the molding of our national character. It may 


well be that a half century from now we shall conclude that the social 
significance of science has less to do with the products that it has inspired, 
with the control of the material world which it has placed at our command, 
with the physical power that it has aided us to subsume from nature, 
important as these are, than with its inner and guiding philosophy ; with an 
ethos involving an overriding personal commitment to enduring values and 
a high personal discipline in fulfilling them. 

In a historical and evolutionary sense, science lies at the very heart of 
western society. Both were born in the common philosophy of the Enlight- 
enment. From the beginning they shared basic principles. For our wider 
society, as for science, the search for truth is an overwhelming value, 
however that truth may be defined. From that commitment to the search 
for truth flow our concern for originality, our maintenance of an attitude of 
skepticism, our protection of and our heavy reliance on the processes of 
pluralism in all we undertake, our defense of argument and dissent. These 
are the values that give to our society and maintain within it the continuing 
power to evolve. 

Perhaps most significant of all, this ethic of the search for truth inherently 
gives special honor to the place of the inquiring individual. Historically, 
we have always depended on him heavily to preserve and implement our 
values, and as a principal bearer of this vital social responsibility we have 
esteemed him accordingly and have consistently protected his freedom and 
provided for his continuing flexibility of action. So if one were to choose 
among the common themes that at a deeper level bind together these four 
disparate developments of the year, the most fundamental would surely be 
that of the individual, and especially the individual deeply concerned with 
science, whether in a strictly professional role or as guardian or implementer 
or observer. What is his place in a scientific world so different in its substan- 
tive scope and richness and complexity from that of his grandfather as 
hardly to be recognizable, and so different in its social impact from that 
even of his father as scarcely to be the same in kind? Is he still truly 
relevant? Or have his functions now been subsumed collectively by the 
great teams and the great organizations of our time, scientific and social 
alike? If he still is truly relevant and vital, is he being adequately prepared, 
today, to comprehend the newer aspects of his multiple roles? Is he being 
trained in sufficient numbers to properly serve an extended and immensely 
evolved science, serving in turn, in a multitude of ways, a vastly expanded 
and elaborated society? 

In a recent and penetrating essay on the emergence of Darwinism, 
commemorating the centenary of the concept of evolution, Sir Julian 


Huxley posed a question of striking concern for our time. What would have 
happened to Darwin as an individual scientist if he had been born a century 
earlier, or a century later? The essayist speculates that if he had been born 
in 1709 he might have gone down to posterity as a sound, and perhaps a 
brilliant, amateur naturalist, possibly somewhat like his grandfather 
Erasmus, whose own philosophy, however, Darwin so strongly rejected. 
It is easy to forget that Erasmus was an ingenious and even a brilliant 
thinker. Yet his notions never really stimulated the development of modern 
concepts of evolution — indeed, they may actually have retarded it. Did this 
reflect on his innate capacities? Or was the failure instead akin to that of 
Edward Blyth, who a generation later, but still well in advance of Darwin's 
publication of the Origin, evolved a concept of natural selection that was 
essentially correct, yet used it to reinforce the idea of the fixity of species 
that was still a dominant concept of the time? Was the overarching struc- 
ture of scientific thought simply not fitted, in Erasmus Darwin's day, to 
accommodate a modern theory of evolution? 

A Charles Darwin born two centuries later, Huxley surmises, would today 
at fifty-five be a respectable, indeed, perhaps an eminent, contemporary 
ecologist, but little more. As a modern disciple of evolutionary thinking, 
conceived as laboring with the same methods and orientation as the original 
Darwin but deep in the shadow of his achievements — or, if the original 
Darwin had never existed, then in the shadow of some near-contemporary 
who inevitably would have proposed the theory — this hypothetical modern 
Darwin would hardly be a revolutionary world figure in today's structure 
of scientific thought. Earlier, the time would have been unripe for his actual 
greatness; later, it is far overripe. 

Few speculations indeed can be more fascinating than this of the probable 
career of a Darwin displaced forward or backward by a century. And few 
could be more relevant to our own time. For it emphasizes strikingly how 
rapidly the whole conceptual structure of even one sector of the scientific 
environment has changed within a hundred years, and how important and 
how different this environmental challenge may be to the individual of 
Darwinian quality and kind who would enter it today. It also poses several 
penetrating questions. When a field of thought has become both far ad- 
vanced and highly elaborated, are its new frontiers still within the reach of 
individuals, however gifted? And if they are, what is the relative significance, 
in the minds of those individuals, of innate quality and of environmental 
influence? Would Darwin's specific aptitudes, his specific fields of interest, 
his specific modes of approaching the world, have persisted more or less 
intact had he been born in 1709 or 1909? If so, were there research questions 
truly relevant to those talents great enough and near enough the surface in 
the two hypothetical life periods to have engaged his interest and proved 
his genius? Is it possible that a mind of such lofty stature, and gifted with 


such an immense capacity for plumbing the underlying verities of a very 
special sector of the natural world, was by those very circumstances deeply 
and irrevocably committed to the arena where it chose in fact to work? 
Was the actual timing of Darwin's life, then, one of those rare near-miracles? 

On the other hand, would a mind of such native originality and power, 
given the appropriate training and opportunity, have chosen quite different 
areas of inquiry in the different periods — areas more accessible in the one 
case, areas involving new unanswered questions of comparable magnitude 
and novelty in the other? Is it not true that in any age of science issues of 
the power and the generality with which Darwin dealt lie below the surface, 
too deep perhaps for ordinary men to perceive, but still accessible to 
individuals of Darwinian penetration, still ready to be detected and revealed 
by such vision — as indeed is happening in adjacent areas of biology at 
present? Would Darwin's particular kind of genius burn as bright at any 
period in the growth of scientific thought about the natural world? Or must 
new kinds of scientific workers, of ever-changing tastes and talents, be 
available in each generation to man the frontiers of a constantly changing 
and elaborating science? 

Some of these questions are clearly unanswerable in the present state of 
our knowledge. Some may not even be real. Yet they bring home with force 
the profound issue of the nature of the influence that existing scientific 
environments have exercised in each generation on the whole scientific 
career of those who have entered such environments most sensitively and 
powerfully. And they underline the unprecedented quality of that environ- 
mental challenge in our own day and in the future, posed equally by the 
complexity of the vast, interconnected, organism-like webs of theory and 
fact, of concept and projection, that collectively make up the vast texture 
of an age of advanced scientific scholarship, on the one hand, and of the 
very different but equally complex organizational environment embracing 
the formal structuring and implementing of science in its relationships to 
the greater society that it serves, on the other. The sheer magnitude, 
the intricacy, the power, and the pervasiveness of those two environments 
in our time must profoundly condition the philosophy and the orientation 
of every individual entering upon the scientific way. Moreover, in their 
headlong future evolution, these environments must in the nature of things 
constantly place new and ever more importunate demands upon him — 
demands that occasionally, in a very general fashion, we can foresee, 
especially in the area of scientific scholarship, but of which, often enough, 
we have little inkling. 

Many challenges to our understanding and action are limned here with 
special clarity. A particularly important one concerns the quality of this 
relationship between native intellectual orientation and the capacity for 
sensitive and discerning reaction to the contemporary scientific environment 


in molding the effectiveness of the young research worker. Closely related is 
another, equally arresting, question. How can the young individual of 
potentially great stature, entering the complex scientific environment of 
today, best be prepared to seek and to uncover and to pursue the issues that 
tomorrow may have major significance? How can he or she learn to recog- 
nize them amid the random noise of the vast and varied contemporary 
scientific scene? 

At least two points of major relevance are here emphasized. The first is 
the cardinal necessity for breadth and flexibility and a fundamental quality 
in preparation — breadth and depth and generality in substance to minimize 
the perils for the entrant too specially trained that can inhere in the rapid 
exhaustion of specific fields so characteristic of modern research; flexibility 
of outlook so that the multifarious relationships of science to the larger 
society will not be foreign to him in principle. The second point is involved 
with that curious and elusive quality perhaps best called the power of 
penetration — the Darwin-like quality displayed by true scientific genius in 
every generation — the power to discern key issues too securely hidden to be 
detected by ordinary men and to generalize from them broadly and effec- 
tively. An important prerequisite for this quality, as we have long recog- 
nized, is the ability to keep many things in mind at once, to experiment and 
play with a multitude of permutations and combinations. This power to 
retain many things simultaneously at the forefront of attention has always 
been an outstanding attribute of individual genius in securing scientific 
advance. In the future, it must be taxed ever more severely as the reach 
required to generate novel insights at the frontiers of many fields inevitably 
increases, as the challenges to significant generalization grow. Will the time 
ever come when the number of elements which must be held and considered 
simultaneously in order to achieve truly great new insights exceeds the 
capacity of even the greatest single mind? Have we any effective working 
tools to increase our powers of penetration further while retaining the 
freedom and flexibility of the individual? Or, in a time of massive science, 
highly organized both within its own complex structure and in its even 
more complex relationships to the rest of our society, has the role of the 
individual actually lost the significance that it had a generation ago? Is the 
challenge too complex to be met effectively except by massive human 

In its every aspect, a continuing exploration of the role of the individual, 
of his tasks, of his preparation, in the changing scientific environments of 
our time, is of far more vital concern to us than ever before in our history. 
For we live today with the very real danger that the special power inhering 
in that role, its unique relation to creative effort in whatever setting, could 
easily be lost in a perspective, all too compelling, which could confuse the 
nature and purposes of organization with actual processes of advance. In 



such an exploration a first concern must be for those environments them- 
selves — the environment provided by the structure and discipline of 
substantive science; the related but yet different environment of science in 
its role of public servant. 

When scientific scholarship is compared with scholarship in other fields, 
or with the arts, it sometimes appears to be distinguished by an important 
difference of kind. It seems to be additive in nature, to grow as the sum of 
facts accumulated by a multitude of workers and piled up from generation 
to generation like the fabrication, brick by brick, of some massive wall. 
Other kinds of scholarship, and the arts, do not seem to be cumulative in 
precisely the same sense. This distinction, indeed, has an element of validity. 
As discoverer, collector, and arranger of new facts about the world, science 
is indeed an activity of accumulation — sometimes even of simple summa- 
tion. And when one looks at the explosive rate at which that activity has 
accelerated in our own nation and over much of the globe since the second 
world war, it is tempting to regard accumulation as a preeminent charac- 
teristic of scientific endeavor. 

Many factors, of course, have contributed to this explosion in the un- 
earthing of new data about our universe. The very storehouse of facts 
already won can accelerate exponentially the uncovering of new facts in 
contiguous areas, so long as those areas themselves remain potentially 
fertile for further discovery. Again, the range and power of the highly 
sophisticated scientific tools developed so conspicuously in this generation — 
and often employed to illumine fields far distant from those for which they 
were originally designed — constitute a most significant factor. Electron 
microscopes and electron probes and techniques of gas-liquid chromatog- 
raphy have opened new areas not only in physics and geology and geo- 
physics. They have nearly revolutionized some aspects of quantitative and 
molecular biology as well. Spectrophotometric techniques of interest to the 
astronomer are also of interest to the student of photosynthesis in green 
plants. Means for measuring the flows of small currents with minute 
electrodes and of amplifying and displaying the results which are vital to 
parts of electronics have also proved powerful tools in investigating phe- 
nomena of the living brain. Advances in the sensitivity of photographic 
plates have been of profound importance in extending the reach of the great 
telescopes of the world. 

Finally, the rapid increase in the numbers of those trained in science and 
active in scientific research in the last decades, though actually inadequate 
to growing opportunities, has been an extremely weighty factor in this 
context, especially in our own country. In 1940, for example, approximately 


one and one-half per cent of the working population of the United States 
were classified as technical specialists. Since then, according to a number of 
demographers, the population of the nation as a whole has increased at an 
annual rate also of about one and one-half per cent. But during the same 
time the numbers of those engaged in scientific and technical work have 
been expanding at an annual rate of roughly five per cent. By 1970, it is 
estimated, some four and seven-tenths per cent of all those employed in the 
country may be comprehended in such specialist occupations. Small wonder 
that the rate of accumulation of scientific and technical facts should proceed 
at an explosive rate, or that the additive property of science should impress 
us so greatly ! 

But if this additive and cumulative aspect of scientific scholarship were 
a true measure of its inner structure, the concepts of our universe would 
have changed in no such dynamic way as they have even over the last two 
decades. A hypothetical Darwin born in 1709, or reincarnated in 1909, 
would have encountered relatively familiar scientific environments instead 
of ones that, as we must imagine, would have been or would now be very 
strange indeed. And if the collection of facts randomly unearthed were the 
only, or even a dominating, factor in the evolution of scientific scholarship, 
there might well be no such pressure as now exists to further expand the 
numbers of scientists in our society. 

For, in a profound sense, the structure of science is no more that of an 
assemblage of facts, brought together in simple additive arrays, than is the 
living body a simple assemblage of cells, coexisting without interaction or 
integration. Rather, like a living body, the body of scientific scholarship is 
a highly organized entity. Its multifarious parts, to be sure, are sometimes 
so diverse and so separated in their various specializations that it is often 
difficult to recognize them even as components of any organized whole. 
Yet so long as the parts are developing dynamically, the essence of an 
integrated relationship is maintained through the whole vast structure of 
science. The intensity or the importance of the reaction of one part of the 
body on another seemingly far distant can never be foreseen, from one year 
to the next or even, sometimes, literally from day to day. But it is a perma- 
nent potential of health and vigor. 

Furthermore, this interlocked character of scientific scholarship, across 
fields and over generations of workers, makes it far more than simply a 
static organic whole. In a very real sense it is a living and evolving organism. 
Its growth over three centuries has been marked, as in so much of actual 
organic evolution, by movement from the simple to the vastly more complex 
and at the same time by a correlated knitting and integration, transforming 
initially loose assemblages of hypothesis and theory and fact into more 
tightly woven, more inclusive, more efficient — and often superficially more 
simple — tools to achieve new orders of understanding. 


As in organic evolution, moreover, the evolution of science is irreversible. 
In any literal sense, scientific scholarship can never go back. One thing must 
follow another: it cannot precede it. Without the work of Newton, the work 
of Einstein would have been impossible ; even if it had been accomplished, 
it would have been irrelevant to the stream of our understanding. Without 
Gregor Mendel's demonstration of the individuality and the integrity of 
the elements of inheritance, the demonstration of the linear arrangements 
of genes in chromosomes by Morgan and his school, three-quarters of a 
century later, even had it been possible, could have had little meaning for 
us. Without Robert Brown's recognition of the nucleus and the cytoplasm 
of the cell in 1831, without the proof of the physical existence of chromo- 
somes in the nucleus by Anton Schneider in 1870, neither the work of 
Mendel nor that of Morgan would have carried the meaning that it did. 
Without all these discoveries and the characterization of "nuclein" by the 
Swedish biochemist Miescher almost contemporaneously with the findings 
of Mendel, without modern developments in physical chemistry and 
information theory and much else besides, the notions that have revolution- 
ized our thinking about the molecular nature of heredity over the last 
decade would surely have been unattainable. Modern physical theories of 
solids could never have been developed apart from the work of molecular 
chemists accomplished two or three decades earlier. And this work in turn 
could not have been accomplished without the conceptual structures of the 
quantum theory, yet a decade older. 

As it has been in the past, so, clearly, will it continue in the future. In 
many new areas it is possible to anticipate even now the outlines of future 
conceptions. A deeper comprehension of the behavior of gas plasmas may 
prove directly relevant to a fuller understanding of conditions in inter- 
planetary space. And as the structure of the interplanetary medium becomes 
clearer, concepts about the outer layers of the solar atmosphere will be 
elaborated and refined. These in turn must bear heavily on our understand- 
ing of the energy balances within our own terrestrial system. 

There will be many instances not only where specific information gained 
in one area will prove directly relevant in very different ones, but where 
major conceptual structures too will be found to underlie fields apparently 
very disparate. Concepts of evolution today illumine the natural world far 
beyond the aspects for which they were originally designed, coloring our 
views in fields as remote as the formation of the elements or the transforma- 
tions of stars. But here, particularly, careful and critical judgments of those 
most capable to take them are vital to the sound extension of any conceptual 
system beyond the boundaries for which it was designed, even though the 
extension in itself may seem quite trivial. The vast advances in our under- 
standing of the genetic code which the decade has brought will surely invite 
us to apply those hard-won notions in other fields where they may or may 


not prove relevant. The visualizing of mental processes in molecular terms, 
for instance, clearly offers one of the most exciting challenges, and one of 
the greatest areas of opportunity, for a new generation of investigators. 
We shall certainly be tempted to think of the mysterious processes of 
memory at the molecular level in terms analogous to those in which we 
have come to conceive the recording and the operation of the information 
of heredity; in terms perhaps of a ribonucleic acid-mediated mechanism, 
sensitively affected by environmental influences acting on the nerve cell. 
Whether any such conceptual bridge can be valid, given the durable and 
highly invariant nature of the genetic mechanism as we now know it, is a 
vital question which in the future must demand one of those key judgments 
so crucial to scientific scholarship. 

As in organic evolution again, the tempo of scientific progress is highly 
uneven, and also varies greatly from place to place within the framework of 
science. Periods of gradual and hard-won advance may be punctuated by 
occasional swift bursts when new insights accomplish rapid and major 
transformations. The general profile of the growth of scientific scholarship, 
indeed, may in one sense be likened to that of beads on a chain. 

Finally, once again as with organic evolution, the overall tempo of the 
evolution of science accelerates as evolution itself proceeds. This combina- 
tion of increase in magnitude and acceleration in tempo has confronted each 
scientific generation with an ever more severe challenge. When Gregor 
Mendel's great work was recovered in 1900 there was a vastly insufficient 
reserve of those whose talents and taste and training equipped them to 
exploit quickly and to the full the advance for the new opportunities that 
it offered. Since then, the number of first-class minds in the world that 
have been oriented and trained in the science of genetics has vastly increased. 
Yet they may still be quite inadequate to take advantage of all the scientific 
possibilities laid bare by, say, recent concepts of genetic coding. In this 
sense there can be no doubt that each generation of research workers has 
been provided with a far richer scientific spectrum than its predecessors. 
This has been true of our own generation to an extraordinary degree. But 
by the same token, each generation has also had to reconcile itself to living 
with an ever-growing backlog of potentials unexploited. 

This situation, too, has clearly reached unprecedented proportions in our 
own time. It carries at least three implications of signal importance. The 
first is that, despite (or precisely because of) the great numbers of those 
already working in scientific fields, we shall always feel that we require 
more, not only in practical terms of national needs, but in terms of the 
growth and vigor of scientific scholarship itself. Yet, just as we must 
constantly beware of attributing to organization per se properties that in 
the last analysis belong to creative individuals, so we must be constantly 
wary of the temptation to imagine that mere numbers of individuals engaged 


in scientific pursuits can in any real sense replace or even compensate for 
lack in individual endowment and quality of training in that smaller corps 
upon which the greatest part of real progress depends. A single genius, we 
must always remember, may accelerate the evolution of his branch of 
science more, in his generation, than a thousand less outstanding con- 

Our enthusiasm for numbers, too, must be tempered on the broader front 
by the continuing recognition that a primary requisite for a healthy society 
is the maintenance of some balance in the proportions of those committed 
to the various functions that serve it. Gross departures from that balance 
can pose a serious threat to social well-being. So in any period there must 
be a limit to the numbers of those who can or ought to be dedicated to 
the attainment of scientific excellence, and we should be continuingly 
sensitive as to approximately what, at any given time, that limit should be. 
Fortunately, however, there is every indication that an appropriate limit 
is actually set by the natural rareness of those who by high talent and by 
burning inclination are destined to become dedicated scientists of the 
highest caliber. So there is little danger that, in terms of such men, we shall 
exceed prudent boundaries in the foreseeable future. Rather, our society 
will continue to be short of them. Nothing, therefore, can come higher on 
our agenda than both the excellence and the extent of the training of those 
among us who in future may choose the scientific way. 

A second implication of the accelerating tendency of scientific evolution, 
though related to the first, has somewhat different practical connotations. 
Since it has never been possible to explore adequately more than a minute 
fraction of the promising leads that science has laid bare at any time, 
careful and rigorous choice among many possibilities has always been a 
characteristic and critical duty of the investigator. Just as in living nature, 
what we see in the evolution of scientific scholarship is in fact the result of 
an uncountable multitude of individual choices of the past, made among a 
plethora of alternatives, each winnowed and scrutinized, each put to the 
hard test of survival. Thus are linked two of the three themes uniting the 
disparate developments with which we began — choice and the individual. 
We shall shortly return to them and to their connection with the third — 
choice and the individual and planning, both in and for the scientific 
endeavor — a process of planning which, as we shall see, carries a very 
different meaning in the two contexts. 

The third implication involves an obverse to this picture of dynamic 
growth. If there are always sectors of science that are rapidly growing and 
evolving, there are also, inevitably, always points of stagnation, always 
branches of investigation that are dying. Few research scientists today 
would choose to enter the field of the determination of atomic weights even 
though, conceivably, greater accuracies could still be obtained. The study 


of stable isotopes of the elements, dynamic in the first half of the century, 
is no longer a very active pursuit. Three decades ago studies of cosmic rays, 
as received at the surface of the earth, provided one of the most exciting 
challenges in all of physics. Today, students trained specifically in that area 
have at times had to turn to other pursuits, while the most exciting growing 
point of the field, studies of the celestial origins of the rays, has become 
the province of astronomy and astrophysics in its widest reaches. Whole 
areas of classificatory biology, and even of more conventional physiology, 
have been exhausted, and there is little chance, in many of them, that a 
revival of interest or indeed of opportunity will come. 

And just as the opening of new opportunities in science is ever accel- 
erating, so too does the tempo at which opportunities are exhausted increase. 
Only two years ago the spectacular discovery was made that xenon, 
traditionally thought to be so unreactive that it has long been known as a 
noble gas, can be made to react readily with fluorine to form xenon tetra- 
fluoride. This dramatic finding opened a new chemical field of considerable 
dimensions. Yet so immediately and intensively was it attacked by capable 
investigators that already the field has been largely exhausted. Over the 
whole front of science, more than one student has found on completing the 
work for his doctoral degree that the very field of his thesis has literally 
disappeared while he was writing it, possibly leaving him with a frightening 
dilemma of career. And the menace is not lessened when the career itself 
has been well launched. The modern development of the computer has 
carried no more serious implications for the untrained worker than it has 
for the intensively but narrowly trained professional numerical calculator 
of the old school — if anything, indeed, it has been less devastating for the 
untrained man, who at least is generalized. The implications for breadth 
and flexibility in scientific training, for maintenance of the fundamentals, 
for the perils of too great concentration too early on too narrowly specialized 
arenas in the preparation of the individual who is to enter science today are 
all too clear. Yet even today they are not always heeded. 

Thus the scientific environment per se. What of the individual and his 
role of judgment and choice in the forming of it? The organic quality of 
scientific scholarship is of course prima facie evidence that the essence of 
science, its shaping and its growth, He in processes of effective integration 
and meaningful organization of information that no mechanical aid yet 
devised and no massive effort yet known are able to accomplish. For they 
are deeply integral with the attainment of insights which both make our 
current vision of the world more comprehensive and can serve as predictors 
beyond the boundaries of present concepts; insights that can prepare our 
understanding for new adventures into the unknown. The attaining of those 


insights, and their validation, must indeed continue to be the work of the 
scientific scholar, so long as we can see into the future. 

Above all, then, the primary motif of a dynamic science is that of constant 
and critical judgment — of making constant and sensitive choices — choices 
among avenues of investigation that are best pursued, choices of direction 
of effort, choices, above all, in the formulation of new hypotheses and in 
judging the relationships among older ones and their soundness and rele- 
vance and applicability outside the fields for which they were designed. 
These will, in general, be choices that are individual and often highly 
specific. Each choice as it is made will be exposed to a constant scrutiny, to 
a constant validation or rejection, to a constant judgment by the facts 
themselves and by a trained and sophisticated audience of co-workers in 
the field. In this pattern of achieving growth as the product of constant 
individual choice, science, far from being distinguished from artistic 
activities in the more usual sense, indeed finds common ground with them. 

This process of choice through which the evolution of science takes place 
constitutes a very real process of planning — none the less real for the fact 
that it is the work of a host of contributors dispersed all over the world and 
through many generations. It is worth emphasis that planning in this sense 
veritably lies at the heart of scientific scholarship. Yet it is very different 
from planning as we ordinarily understand that word. Its highly contingent 
nature makes of it a very special kind of planning. It is planning at the 
level of the work, and its whole effectiveness in the past has rested on the 
excellence of judgment and the freedom of judgment of individuals whose 
abilities and preparation fitted them to make the specific choices demanded, 
and also to judge similar choices made by colleagues. The words of Elihu 
Root, written in 1918, carry a lasting and special significance: "Occasionally 
a man appears who has the instinct to reject the negligible. A very great 
mind goes directly to the decisive fact, the determining symptom, and 
cannot afford to burden itself with a great mass of unimportant facts. ..." 
It is such minds, acting with individual freedom and flexibility, that provide 
the sinews of our scientific planning today, as they did a half century ago, 
and as they surely will into an indefinite future. It is not, and it inherently 
cannot be, the function of the great organization per se, or of planning as 
ordinarily understood. 

But if the investigating mind should not be burdened with a mass of 
unimportant facts, it still cannot discover or test potentially significant 
relationships without taking into account a plethora of collateral evidence 
which, though less salient to the principal lines of reasoning, is yet con- 
tributory to it. Consciously or not, the individual pioneer must have access 
to a wide range of facts and hypotheses as he ponders new combinations. 
Outstanding ability to hold such a store in mind is indeed one frequent 
attribute of genius, and is commonly a prerequisite to original and pene- 
trating thought. Today, with the frontiers of science become in many areas 


so wide and many-faceted, this individual ability in even its highest 
expression has become hard pressed. In some areas it may even have become 
limiting. And so we return to that insistent and urgent question of whether 
there are possibilities of structuring the scientific effort to provide the 
pioneering individual with more powerful investigative tools without at the 
same time submerging his identity or limiting his initiative. 

There are such means, and in principle they have had a long development. 
The inherent limitations posed to even the greatest minds by the vastness 
of the subject material with which they must deal are surely part of the 
reason why individuals in science have always, and characteristically, 
sought to form unusually intimate working communities, tightly con- 
structed about their disciplines, however widely physically separated their 
members might be. In our own day, such communities have on occasion 
evolved further, to form deeply intimate associations, working so closely 
over long periods on matters of common concern that the group may 
ultimately become truly greater and more powerful than the sum of its 
parts — truly a ''thinking organism" in its own right. In many fields such 
intellectually close-knit, concentrated teams have become identified with 
some of the most significant conceptual advances in the science of our time, 
comparable to and often reminiscent of the kinds of advances achieved in 
another age by the individual genius of a Heaviside or a Pasteur — such 
groups as those within the Pasteur Institute itself, of the Lawrence Radia- 
tion Laboratory, of the Bell Telephone Laboratories, of the British Medical 
Research Council's Laboratory of Molecular Biology at Cambridge, of the 
Carnegie Institution, to cite but a representative handful. Indeed, it is 
possible that if one looks to those larger and apparently unitary organiza- 
tions where scientific frontiers are being most spectacularly pushed back 
today, and examines them in close detail — in industry or government, 
within private research institutions and even to a considerable extent within 
universities — it will be found that in a great number of instances those 
conquests depend in the last analysis upon the work of such relatively small, 
close-knit, and concentrated teams harbored within them. 

A particularly powerful quality of teams of this kind lies precisely in 
their compatibility with the play of individual genius. It seems clear that in 
the ideal research group of this sort not only are the unfettered freedom 
and individuality of each of its key members carefully guarded as a factor 
vital to success but also a major function of the group is indeed to provide 
an environment where such individuality will be enhanced and stimulated 
to the utmost, fired by common concern and common sympathy, given 
access to a stream of ideas and a collection of facts broader and deeper than 
any individual alone could reach, exposed to resources of criticism far more 
apposite and stringent than could be experienced in solitude. Such groups 
may indeed offer particularly relevant and powerful responses for our time 
to the encompassing issue of how we are to devise effective means further 


to increase our powers of penetration into the unknown, and of the place 
of the independent investigator in them. 

By all the evidence, then, it seems abundantly clear that the role of the 
individual in the processes of very special substantive planning through 
which science advances, of the individual whether working alone or in small, 
unfettered teams, continues as indispensable in our own time as it has ever 
been. Its importance needs repeated emphasis. For it is easier than ever 
today to forget this cardinal truth, or to lose it in the obscuring image of the 
giant scientific and technical organizations with which we live. It is natural, 
but gravely misleading, to regard such organizations as monolithic entities, 
failing to discern in their finer structure the congeries of close-knit groups 
of individual investigators that bring them their vitality and effectiveness 
in research. 

Such misconceptions may be perilous. They could exacerbate a trend, 
already evident and probably destined to become more conspicuous, that 
can easily dominate the climate of thought and feeling in a technically 
advanced society — a trend toward a progressive secularization of the search 
for new knowledge, and especially of knowledge related to scientific and 
technical discovery: the taking-for-granted both of knowledge won and of 
the processes of winning new knowledge that is so apt to cloud our horizons. 
What happens to the emotionally compelling quality of great new conquests 
of nature, of great new technical developments, of comprehensive new 
systems of ideas, as the impersonal concerns of large-scale integration 
displace those of discovery itself, with its early wonder? What may happen, 
not only to quality, but also to standards of quality? How often may the 
temptation that is no stranger to us today seem wholly logical and even 
efficient, the temptation to substitute money or sheer volume of numbers 
for hard thought, often enough not in sophistry but in what is actually 
much worse, honest self-deception of a peculiarly insidious kind? 

Such dangers pose important challenges indeed. For it may be deeply 
characteristic of the human mind that, as the organization and the system- 
atization of knowledge proceed, the color and the vividness that surround 
the acquisition of new knowledge become progressively more difficult to 
maintain. Yet it is just at this point in our history that those elements 
become most vitally important. Verve and style and deep individual 
involvement are indeed inseparable from quality, and from the adequate 
making and the adequate monitoring of those perpetual and myriad 
individual choices upon which the progress and the final shape of science 
ultimately depend. The significance of their maintenance must become 
increasingly pressing with the years. By the standards of training and of 
work, through the great unities of approach and of preparation that bind 
those dedicated to the scientific path : the requirements of verifiability ; the 
discipline of parsimony; the constant emphasis on the significance of 
individual effort with its exacting demands of originality and imagination, 


of the maintenance of style, the individual scientist, so long as he remains 
conscious of the weight and seriousness of his personal responsibility, can 
do much to preserve those priceless qualities, so often confused and 
threatened with destruction in a contemporary crowded world. No era to 
which they have been lost can be great, whatever its other assets. The 
responsibility to conserve them is a major one indeed. And it belongs 
supremely to this generation. 

So much for the processes of substantive scientific planning, for planning 
in science and its relation to choice and to the ultimate importance of the 
individual. What then of the different question of planning for science, in 
its effects on and its relations to our greater society? With science as 
substantively immense and intricate as it now is, with science as demanding 
as it has become in terms of the material and the human wealth of nations, 
with science, in short, so overwhelming a social force on both intellectual 
and material fronts, can we longer say that we can allow it to proceed 
without a measure of planning from without? And, if we must affirm this, 
what do we mean? Is the process of planning for science remotely like that of 
planning in science? Is planning for science, like that of planning in science, 
intimately linked still in our day with the work of individuals, or is it, in 
its vastness, subject only to the influence of relatively impersonal organi- 

Were a Darwin, miraculously reborn in 1909, observing us today, he 
would find the landscape of science vastly altered from what he knew in 
features quite other than the richness and the diversity, the sophistication 
and the complexity of its inner substance. One touchstone of its accom- 
panying outer transformation would surely be the vast increase in numbers 
of those engaged in it. Another, and one yet more poignant in this context, 
would be the spectacular growth in public expenditures in scientific and 
technical fields. Among all the industrial nations of the world that increase 
has been explosive. It has been especially vividly illustrated in our own 
country during the postwar period. In 1950, for example, our national 
expenditures for research and development amounted to about 3 billion 
dollars. By 1960 they had grown to 13 billion, averaging more than a 
doubling for each five years. In the current year the appropriation exceeded 
15 billion. Such a rate of increase is to be set against a rise of only about 3.5 
per cent in the gross national product of the nation. If we were to project 
both rates unchanged to 1973 — an unlikely hypothesis but one underlining 
strongly the drama of the situation — our national expenditures for research 
and development would comprehend no less than 10 per cent of our gross 
national product. 

And the share of this burden carried by the federal government has grown 


formidably in the last years. In 1940, the federal government spent about 
$74 million in research and development. By 1960 it supplied about 10 
billion dollars, as against $4.7 billion provided by private industry, while 
colleges, universities, and private scientific institutes, still probably the 
primary producers of research, supported only about 3 per cent of the total. 
By 1963, when the total expenditure for research and development in the 
nation had reached almost its present level, the government had come to 
support more than two-thirds of it, at an annual rate greater than the total 
sums it had committed to research and development from the time of the 
American Revolution to and through World War I. It comprehended 15 
per cent of the total federal budget, and about one-third of the part not 
already absorbed in fixed commitments. 

At fifty-five, a reborn Darwin would indeed work within a science which, 
over all the industrialized world, had changed greatly in the character of its 
relationships to societies and to governments. To be sure, the original 
Darwin must have been very familiar with the beginnings of organization 
at the interfaces of science and of society at large. He was just twenty years 
old when James Smithson died, leaving his bequest of half a million dollars 
for the further promotion of American science. He was thirty-nine when the 
American Association for the Advancement of Science was organized in 
Philadelphia. It is easy to imagine how familiar in type such private 
organizations designed to promote the service of science to society would 
have seemed to him, and how much they might have recalled some of the 
contemporary and parallel work of corresponding bodies in Great Britain. 
He was fifty-four when the National Academy of Sciences was organized, 
and one can imagine how comprehensible such a systematized organization 
of science in the public service would have been to him, and how logical it 
would have seemed. 

But within three-quarters of a century after Darwin's death three 
important developments had occurred to fashion in the social landscape of 
science a topography so different as to be, in essence, new in kind. The first 
has the longest history, and has always been the most obvious. The pre- 
cipitate rise of industrial power, and the critical place of both science and 
technology in evolving and maintaining it, were ultimately to bring to both 
a new kind of social significance. The second development, which has 
dominated the last quarter of a century, was the recognition brought by 
three great wars of the indispensable role that science must play in defense 
and in the maintenance of military capacity. The third, taking its root in 
the first two, is embodied in the final conviction of our day that contempo- 
rary science and its associated technology have become in themselves major 
forces changing the very face of society. Associated with all these develop- 
ments, but most particularly with the last, has come an increase in the 
public expenditures for science in many lands to the point where they 
constitute major items in overall national budgets. 


These transformations of position and of attitude, subtle but surely 
profound, have brought yet another in their train. With science of such an 
order of importance for societies and for nations, with science supported at 
such a level of public expenditure, science must henceforth be as directly 
accountable to its supporting public as any other activity of like social 
impact. In this context, science is a public activity of major importance — 
public and major, and so, inevitably, political. And in this context it may 
not be generally visualized as the private exercise of great intellects in the 
exploration of the universe. It may not, indeed, be generally thought of as 
an activity of individuals at all. 

This orientation would surely have seemed very unfamiliar to Darwin, 
though its beginnings in our own country can certainly be traced at least as 
far back as his lifetime, in the philosophy and the organization and the 
conduct of the Department of Agriculture itself. But, spurred by the great 
shifts in the center of gravity of the large-scale practical implications of 
technology from agriculture to such immediate life-and-death matters as 
national defense and military power, and to the equally vital arena of the 
transformations of industrial societies, the change in our own country has 
accelerated with astonishing rapidity. Indeed, the full change has actually 
been surprisingly recent. As late as 1938 one-third of all the federal money 
committed to research and development was expended by the Department 
of Agriculture, and the Departments of Commerce and of the Interior 
combined accounted for approximately another fifth. The entire military 
budget of the Departments of War and of the Navy consumed only a fifth. 
Only fifteen years later, in 1953, nine-tenths of the entire federal budget for 
research and development was to be devoted to military or military- 
oriented research, expended largely through a group of giant federal or 
quasi-federal agencies. Within that period had come the second world war, 
and all that it brought in knowledge of the power of science in national 
defense when appropriately organized and conducted at the federal level. 

All over the globe since World War II the effects of these powerful 
influences to make of science a major object as well as a major instrument 
of public policy have been signalized by the formal organizational relation- 
ships of science to government which have been architected in many nations. 
In the United Kingdom a Ministry of Science and Education was created 
not long ago. And currently, in the Zuckerman Report, the Trend Report, 
the Robbins Report, and in a plethora of public discussions, national issues 
related to this change are being explored with unprecedented range and 
thoroughness. In France a General Delegacy for Scientific and Technical 
Research has been established, directly attached to the office of the Prime 
Minister and supporting an Interdepartmental Committee of Scientific 
Research composed of eight Cabinet Ministers and chaired by the Prime 
Minister. A Consultative Committee on Scientific and Technological Re- 
search has also been set up in France to assist the Interdepartmental 


Committee, consisting of twelve senior professional members, and there is 
also a National Center for Scientific Research. In Belgium there is a 
National Council for Science Policy, in Norway a Government Council for 
Applied Research. And now in Sweden, as noted earlier, a Science Advisory 
Council has very recently been established. In Canada the National 
Research Council and in Australia the Commonwealth Scientific and 
Industrial Research Organization have assumed tasks of new magnitude in 
linking science and national policy. 

Such developments, moreover, are by no means confined to Western 
Europe and North America. Turkey, Iran, and Pakistan already have, or 
are planning, National Science Councils, and Pakistan, among a number of 
bodies concerned with science, has a vigorous Department of Scientific and 
Industrial Research serving a national function quite similar to that long 
served by the department of the same name in Great Britain. Indonesia has 
recently appointed a Minister for Science, and Thailand, with the help of 
Australia, is currently setting out on the same general road. 

When scientific goals and activity assume such proportions in the lives 
of nations as at present, when science has become so significant an element 
in public policy, administrative centralization of the conduct of science 
within government clearly becomes, in one sense, a wave of the future. So 
obviously adaptive is it that all industrial nations are sure to face continuing 
pressures to evolve ever further in that direction. Such trends will inevitably 
be additionally emphasized by the ever-mounting economic pressures that 
a growing public science and technology must continue to impose, further 
increasing the concentration of support and control in central governments. 

Although it is only rather recently that we have felt the full weight of 
this problem, in its essence it has been with us for longer than Darwin's 
lifetime. In a historical sense, we have been exceptionally fortunate in 
meeting it. For not only did our pragmatic American temper secure a firm 
conjunction between science and more general aspects of government policy 
early in our development. It also achieved that participation in a very 
"plural" way, at a variety of levels and through a variety of devices that 
were ultimately to proliferate into an extraordinary panoply of departments, 
bureaus, special working groups, special advisory bodies. Moreover, many 
of the individuals who initiated or early participated in these activities in 
government retained throughout their official service the closest of pro- 
fessional connections with their colleagues in private scientific bodies. 

Throughout our history such circumstances have brought to our conduct 
of science in government a curious and probably unique balance between 
pluralism and centralization. In times of military or economic crisis in the 
nation, when the great issues were intensified and sharpened and simplified, 
we have always tended to concentrate our administrative concerns for 
science at the national level. In times of lesser national stress, with the issues 


of the country far more diffuse, our modes of meeting them have also been 
much more diverse. This quality in our relationships of science to national 
affairs is deeply rooted in a characteristic of our approach to policy forma- 
tion in general which stems in turn from the dominantly pragmatic orienta- 
tion toward affairs that we have had from the very beginning. More than a 
hundred years ago Alexis de Tocqueville, who knew us so preternaturally 
well — often so much better than we knew ourselves — emphasized that an 
especially striking quality of the American ethos was our strong penchant 
for solving the practical problems that we encountered each as it presented 
itself, each as it occurred, in its own way. This strong predilection for 
approaching our problems of policy on an essentially contingent basis has 
persisted through all our history, and is undimmed with us today. It has 
had an especially strong impact upon the whole character of the relation- 
ships of our science and technology with government and its influences on 
governmental policy, and indeed upon the whole cast of our thinking about 
the place of science and technology in our society in the broadest sense. As 
in the progress of science itself, the contingent mode of planning confers a 
special flexibility on all our thinking of the relationships of science to our 
society, and it has allowed us over the years to change and modify our policy 
with unusual swiftness and ease as conditions and demands were changed. 

But we have to remember that this pragmatic style of contingency, when 
overemphasized, can also exact its own price. We are traditionally excellent 
at individual scientific and technical tasks, often of the highest difficulty. 
But we have likewise often been indifferent at broad planning for science. 
Clearly such indifference is no longer admissible. For not only are the 
concerns of science and technology now important objects of national policy. 
Not only are they virtually omnipresent, affecting almost every other aspect 
of our national life. Their very demands, not only of material wealth, but 
especially in terms of the lives and fates of men and women, have become so 
great that, were we to attempt to satisfy all of them indiscriminately, we 
should so overtax both our material and our human resources as to risk 
serious social imbalances. Currents of such magnitude and import as those 
now sweeping science and technology on their courses, if allowed to flow 
undirected, if left to the whim of the winds, could easily wreak great damage. 
So we are irrevocably brought face to face with a grave conclusion. In our 
day, planning for science, though very different from it, may become as 
inescapable a necessity as planning in science has always been. Upon how 
wisely and well we learn to do it, indeed, much of the freedom and vitality 
of substantive science itself will depend. Skillful and effective planning for 
science can enhance the freedom with which active scientists can choose 
and shape the direction of their research. Unskillfully attempted, it is 
capable of restricting that freedom dangerously. 

Now planning for science, too, is clearly accomplished as the sum of a 


multitude of choices, and of choices made by individuals. But how much 
more difficult, and in many ways how very different, these choices often are! 
Not only may the immediate social impact of any single decision be much 
greater. Its effective verification— the possibility of determining whether it 
has actually been good or bad — may be much more distant in time. Circum- 
stances apparently quite foreign to the choice yet actually vital to it may 
change while it is being made. And choices of such magnitude cannot be 
easily altered or rectified if they have initially been unwise. Moreover, 
decisions so important to the whole life of the nation must inevitably be 
taken publicly and be made only with public consent, given or withheld by 
a public who may or may not be prepared to really understand the content 
and to project the consequences of the choices involved. 

This necessary circumstance in turn involves two further hazards. One 
is that many decisions concerning science and technology inevitably must 
critically shape the world in which a subsequent generation — perhaps 
unborn at the time that the decisions were taken — will live and carry on its 
work. How is public choice or public consent to bear effectively upon a 
situation of this kind? A related hazard is that since popular expectations 
of the power of science have grown so immensely in our time, and since such 
expectations must inevitably play a dominant role in governing decision, 
there is the grave danger that when public attention becomes focused upon 
the practical benefits of science, real or potential, anticipation which is 
sometimes unrealistic may father belief, and may, in the absence of wise 
counsel, lead to serious distortions of scientific choice. Spectacular enter- 
prises may find favor at the expense of less conspicuous but basically more 
significant elements; the immediate may take precedence over the long 

It is not only in the weight of decisions taken and in the methods of 
validating and implementing them that planning for science differs from 
planning in science. It can also be far more complex in a substantive sense. 
The farther planning for science in the service of public affairs is removed 
from regions where objective verification is possible, the farther, as it were, 
it is removed from the workbench, the more difficult judgments become, 
and the more unreliable they tend to be if taken in a purely scientific 
context. Indeed, the virtual impossibility of planning large-scale scientific 
or technical enterprises — and particularly of choosing wisely among a group 
of such large-scale enterprises that compete for common and limited sup- 
plies of money and above all of men — on the basis of objective scientific 
considerations alone poses a critical kind of dilemma with which we have 
lately become thoroughly familiar. 

In the processes through which the evolution of the body of scientific 
knowledge takes place, the elements judged at each point of choice can 
commonly be compassed within a relatively straightforward hierarchy of 


values. The construction of such a hierarchy, to be sure, may be a difficult 
process, often requiring as much skill and effort as the final act of judging 
itself. But one of the accepted and understood challenges to the research 
worker is precisely to accomplish such a reduction, however diverse his 
subject material, so that ultimately the values to be weighed are in a real 
sense commensurable. In the more detailed ranges of planning for science 
within a public context it may be possible to preserve this commensurability 
to some degree, and clearly one of the important administrative tasks for 
those dealing with scientific affairs in government at more specific levels 
is this kind of ordering. 

But however much we are aware of this need, and however excellent the 
skills that we develop to accomplish it, judgments made at the highest 
ranges must continue to be taken among elements which, in any basic 
scientific sense, will always be intrinsically incommensurable. How is it 
possible, for example, to determine on a purely objective basis the relative 
priorities in public encouragement and support to be given the field of 
oceanography as against that of mental health, of molecular biology against 
geophysics or atmospheric research, of physiological psychology against 
organic chemistry? At the most obvious level, no observer or judge, regard- 
less of how expert he may be in a particular scientific area, is likely to be 
able to effectively assess in detail the relationships of all the multifarious 
factors involved in fields outside his own or to understand specifically their 
significance for the future. But the real difficulty goes much deeper. For, 
regardless of the extent of scientific knowledge and the wisdom of any 
observer, at this level of decision there can be no single criterion of choice. 
By the very nature of the situation, the orders of merit in such a hierarchy 
of preference must always be plural and often enough inchoate. And they 
will depend in turn upon a broad and complex range of factors. Some factors 
will be principally scientific. But many will be intimately tied to a wide 
variety of other bounding conditions, peculiar to the time and place of the 
decision and varying constantly. 

It is particularly difficult for us to fully recognize and accept the reality 
that there are reaches of decisions involving scientific matters — and often 
among the most important in our day — where the choices to be made must 
remain forever incommensurable. With our long conditioning, we may tend 
to respond negatively to the challenges that this cardinal fact imposes, and 
in ways that can be dangerous. We may deny this inevitable incommen- 
surability, insisting that choices of such nature can always, with sufficient 
effort, be ranked in a single overall " order of merit." Such an error can easily 
result in undue focusing on a few elements of a great problem to the detri- 
ment of the whole, bringing about distortions of judgment of a kind that 
we have not been stranger to in recent years. Another reaction epitomizes 
the opposite extreme: to conclude that, since choice must always be multiple 


and must often be made among elements that cannot be weighed objec- 
tively, planning for science is itself a deceptive and inadmissible activity. 
But we have no option. For the abjuring of planning is in itself a decision 
of the most significant kind, and a potentially disastrous one. We have no 
escape from the conclusion that, the weightier scientific decisions become, 
economically and in affairs, the more they demand judgments of just this 
plural kind, as sensitive to social context as to scientific content. The 
greater the decisions, the more it is incumbent upon those who take them 
to combine with keen scientific judgment the broad, intuitive, socially based 
wisdom that is so universally needed elsewhere in great affairs. 

The whole process through which the entire modern fabric of scientific 
thought has come to being has clearly been the resultant of a plethora of 
individual choices. Its effectiveness has always depended critically on that 
corps of imaginative individuals of the past and the present who, in the 
quality of their training, in the nature of their dedication, in the high order 
of their capacity, have by their collective decisions carried science forward. 
We are beginning now to see that the great future task of planning for 
science, which in some measure we cannot escape, must, in similar fashion, 
depend on the collective choices and judgments of individuals as highly 
qualified. Who will they be? Where will they come from? How will they 
be trained? How shall we recognize them? 

It is abundantly clear that the scientific and technical skills and knowl- 
edge and judgment that are brought to the service of the nation at govern- 
mental level are not only plural in the points at which they enter and the 
avenues through which they serve. They are also, in a deeper sense, multiple 
in kind. They range all the way from those adapted primarily to providing 
the most practical and straightforward technical services to those particu- 
larly qualified to formulate far-ranging policy involving scientific param- 
eters, but including much else besides. It is surely important to draw 
distinctions among these skills and functions, for men and women of very 
different abilities and temperament and training are — and in the future even 
more will be — required to implement the various sectors. 

In the realm of policy alone, one can, by greatly simplifying the picture, 
distinguish at least four general kinds of function. They might be described 
as the execution of policy involving matters of science and technology on 
the domestic front of national affairs ; the corresponding function in relation 
to our foreign policy ; and, on both foreign and domestic fronts, the making 
of those great and often inchoate judgments by which policy in both arenas 
must be formulated. Each of these classes of function is surely sufficiently 
complex in itself. To make matters worse, two or more may often be 


combined in a single demanding situation, as was the case, for example, in 
the extremely difficult problems posed for the members of the Geneva 
Conference in 1958. 

These more or less artificially distinguished levels of function must find 
their counterparts in the kinds of background and training and talents 
required of those called upon to discharge them. Both the formulation and 
the execution of what might be called "general and administrative scientific 
policy," for instance, must surely demand, among other things, constant 
determinations of the distribution of national emphasis among broad sectors 
of the scientific and technical effort. It must include vital considerations of 
the preparation and the distribution of the nation's precious asset of 
scientifically trained men and women. In such a context it must be con- 
cerned with how an adequate cadre can be protected and conserved for the 
work of preparing future trained generations, against the importunate 
demands for immediate service that often bear so heavily today. In every 
sector it will be constantly involved in judging issues similarly related, 
sometimes apparently only distantly, yet actually subtly entwined. This is 
not a " creative" activity of science as that word is commonly understood. 
But it may surely be far more difficult to fulfill than some major "creative" 
tasks. And it entails responsibilities for those involved that are serious and 
major. For upon how well such policy judgments are made may depend 
much of the freedom and the flexibility of creative scientific activity in the 
future. Surely, therefore, due attention must be paid to this capacity for 
judgment among incommensurables in those who will serve on such fron- 
tiers, and some training for wisdom in this kind of choice must be reflected 
in their selection and their preparation. 

Difficult as such judgments are, they do involve the kinds of decisions 
that are regularly undertaken, even though on a less massive and complex 
scale, by able, bold, wide-visioned administrators in industry : the presidents 
of large industrial corporations or their vice presidents for research. Ex- 
tensive and intensive scientific training is of the highest importance for such 
men. But it must be combined with administrative judgment and acumen 
in their most comprehensive sense. Scientific originality in its more strictly 
substantive context may be less essential. Surely we should look importantly 
in the future to our existing reservoirs of such talents and training and 
experience to implement these frontiers of planning for science in the 
national service. 

Most difficult of all the functions, of course, is that field of planning 
where the choice of general political course in a scientific climate is involved. 
This is the field of the individual significantly related at high policy level to 
projecting the development of science at home, or of the drafter of some 
new Truman-King- Attlee declaration, or of the negotiator at some Geneva 
Conference of the future. Those who patrol such frontiers will need to have 


the broadest kind of scientific competence. Further, they should, if at all 
possible, have to their credit personal scientific achievements of significance, 
not only as preparation for comprehending the substantive aspects of the 
questions with which they will deal but also to enable them to command the 
respect and the service of men of science of the caliber who must be their 
advisers. At the same time, they must possess a fundamental understanding 
both of public policy and of the modes and patterns of public representation 
and negotiation. 

Such men and women are rare among us. How rare they have always been 
is vividly evidenced by the cruelly heavy burden that the few have had to 
bear in every major emergency. There may never be a large number in our 
society who are endowed with the capacities demanded in such a role, who 
can through their native gifts and backgrounds assume the responsibilities 
imposed by it. 

But, if we have always been short of such men and women, their paucity 
is uniquely critical today. And it can be expected to become ever more 
critical in the future. Don Price has expressed the requirements with 
extraordinary clarity and vividness : 

What is needed is a corps of men whose liberal education includes an appreciation of 
the role of science and technology in society and whose education has not been a 
narrowly technical or vocational one, but has treated science as one of the highest 
intellectual endeavors of men who also have the responsibility of free citizens. 

Among all the demands of our day in the field of science and national policy, 
the challenge of finding those who are equipped to work in this most 
exacting arena, and of properly preparing them to assume their tasks, stands 
preeminent. In the breadth and balance of taste and preparation between 
scientific matters and wider and more general concerns that it demands, it 
parallels the contemporary need for appropriate balance in taste and 
training between specialized and more generalized subject fields which is 
such a vital requirement in our time for those destined to enter substantive 
science. Both requirements present very special challenges to education in 
its widest and deepest reaches. How well they are acquitted in the years to 
come will determine in no inconsiderable measure the future of America. 

If Charles Darwin stood among us today he would indeed find the 
superficial features of the landscape of science — the landscape that he knew 
and loved so well — immensely changed. In its substance, he would find 
conceptual structures, even in the areas that he knew best, so advanced in 
sophistication, so enormously elaborated, so greatly enriched in new points 
of growth and challenge, so altered in its emphases of old ones, that he 


would surely stand bewildered. And in the broader reaches of the relation- 
ships of science to society he would find a context so altered and enlarged, 
so different in the order of its implications and of the immanent demands 
pressed upon it, as to be virtually new in kind. 

But perhaps he would not long remain wholly puzzled. For permeating 
the evolution of the whole fabric of substantive science, and of the newer 
fabric of planning for science in the large as well, he would surely discern a 
critical and invariant theme that would be thoroughly familiar to him. It 
would indeed be that permanent theme of the work and the essential place 
and function of individuals — dedicated individuals upon whose myriad and 
collective shoulders all those patterns of critical choice have rested and 
must rest which, at the level of planning in science, have preserved sound- 
ness and guided the course of its evolution, and in the province of planning 
for science must in future determine the breadth and the soundness and 
the relevance of its relations to the whole social fabric. Upon that myriad 
corps of individuals, upon the adequacy of their preparation, upon their 
capacity for continued and growing vision, the health and the effectiveness 
of our science must continue to depend absolutely. Today the challenges 
are far more complex, more multifarious, and often even more difficult to 
define, than anything Darwin knew. But surely they do not differ funda- 
mentally in kind. In the requirements of individual preparation and of 
individual dedication that they pose, and in the supreme demand of our 
time for great and qualified individuals in science and in scientific affairs, 
whatever their specific functions and contributions to the whole, a Darwin 
would surely recognize the same overriding and familiar need that domi- 
nated his own age, and to which his own life so magnificently contributed. 

Such functions surely call for capacious minds and reliable powers for disinterested and 
fair-minded judgment. It demands the habit of curbing any tendency to reach results 
agreeable to desire or to embrace the solution of a problem before exhausting its compre- 
hensive analysis. . . .Its task is to seize the permanent, more or less, from the 
feelings and fluctuations of the transient. 

Felix Frankfurter — The Supreme 
Court in the Mirror of Justices 

The Year in Review 

The Republic of Science is a Society of Explorers— Michael Polanyi, Minerva I (1962), 72 

The similarity of scientific research today to the geographical explorations 
of earlier centuries has been described so often that it now belongs to the 
folklore of all advanced countries. The frequency with which the word 
"frontier" is used by scientists is convincing evidence of the impression 
this point of view has made on science itself. In fact, there is a great deal of 
truth in the analogy. Each member of this modern Society of Explorers has 
the sense of commitment and responsibility given by the individual choice 
of objective; he has the stimulation of forming an hypothesis about what 
he hopes to find ; he experiences the excitement and hardships of following 
the unknown trail. Many have the elation of standing on hitherto unknown 
ground, and some the gratification of reaching a goal. There is the same 
freedom, the same loneliness, and also, by contrast, at times an intense 
community feeling with those who have shared the hunt, with its dis- 
appointments and joys. 

The Society of which Professor Polanyi speaks is in general loosely 
organized, but it is characterized by many specialized communities, whose 
judgments are exacting and inescapable, and whose companionship is a 
prized reward to members. 

During 1963-1964 the Carnegie Institution had in residence 200 men and 
women who qualified as explorers on this modern frontier. Sixty-eight of 
them were Staff Members of the Institution. It also was host to 132 others, 
who as guest investigators or fellows spent all the year or some part of it 
pursuing their searches in the laboratories or observatories of the Institu- 
tion. It may be interesting to see how these people conducted their ex- 
plorations, as well as what they did during the year, and what goals they 
reached. The description inevitably touches upon their place in the com- 
munity of science at large, that is, their relations with others in the Society 
of Explorers. 

The Institution contains six research organizations in the biological and 
physical sciences. Each is administered by a Director, who is responsible to 
the President of the Institution for policy and operations in his field. 



Nevertheless, this responsibility is so interpreted that the individual Staff 
Member of the Institution has a high degree of independence. As J. D. 
Ebert, Director of the Department of Embryology, comments in his report 
this year: "Each member of the staff has the opportunity of following his 
own ideas even if they do not fit the latest style. All that is required is his 
conviction that his approach fit his idea of how his subject ought to 
develop." These remarks may be applied to the operational habits of every 
department in the Institution. Independence means the freedom to choose 
an objective, freedom to choose the means, pace, and route in attempting 
to reach the objective, and freedom to choose associates and communicants. 
The manner in which scientists' choices develop, as illustrated by the 
program of the Institution in 1963-1964, is very illuminating as to the way 
of science in this age. Theoretically, a scientific Staff Member of the 
Institution could work in complete isolation if he chose to do so. In practice, 
none does. A number work as individuals, but there are always, even in the 
most esoteric fields, at least a few scientists somewhere else in the world 
working in parallel with them, with whom an individual worker must 
communicate if he is to maintain an up-to-date perspective on his work. 
It is interesting, and perhaps significant, that many scientists in the Insti- 
tution choose to work in groups, either as a team of peers or as an association 
of a senior leader with junior participants. Intradepartmental groups 
carried on work during the year, but there also were interdepartmental 
research associations, and interinstitutional teams collaborating with other 
organizations in the United States as well as with institutions abroad. Some 
chose research problems on which there are very few parallel efforts else- 
where in the world; others chose areas with many parallel research workers. 
Indeed, it might even be said that some of the areas were " highly com- 
petitive." In not a few cases several explorers were attempting to reach the 
same goal by the same route. 

The Mount Wilson and Palomar Observatories 

Most astronomers are literally, as well as figuratively, explorers. Their 
search covers so vast a sweep of space that it is almost impossible for the 
average human imagination to comprehend it. Perhaps in part because of 
this, but also because of the nature of the instruments used, astronomy from 
the time of Hipparchus onward offers many classic examples of the im- 
portance of individual effort. Among all the sciences perhaps only in 
mathematics is the discreteness of the individual mind and its effort seen 
more clearly than in astronomy. This remains true today, as a perusal of 
the astronomy section of recent Year Books of the Institution will readily 

Yet even in astronomy the connections with the " Society" at large are 


important. Dr. H. W. Babcock, the new Director of the Mount Wilson and 
Palomar Observatories, recently said: "No research is isolated. All progress 
is dependent on communication. There is a web of innumerable connections 
between individual minds that work on frontier problems. Some of the 
strands lie in the present; others extend back in to the more solid fabric of 
the past. The connections can be traced in practically any current research 
that one cares to mention." 1 

The ways of modern astronomy and of the Observatories today are well 
illustrated in the year's work on the optical qualities of radio sources, 
investigations that are at the very forefront of present astronomical interest. 
They provided some of the more striking results in the work of the Observa- 
tories during the year. 

Of the five possible ways of observing objects in space, 2 only one, optical 
signals, had been used until a relatively few years ago. It is only during the 
last twenty years that optical astronomy has been supplemented by radio 
astronomy, following the discovery in 1932 by K. G. Jansky, of the Bell 
Telephone Laboratories, that radio waves arrived at the earth from outer 
space. In the years after the end of the Second World War, hundreds of 
radio sources were detected, but lack of precise locational techniques 
prevented the correlation of more than a few sources with the wealth of 
accumulated data from optical astronomy. Within the last few years, 
however, through interferometry and lunar occultation techniques, more 
precise positions have been obtainable for the radio sources, optical identi- 
fication of them became practicable, and the possibility of integrating radio 
data on discrete sources into astronomical knowledge suddenly emerged. 
Within the last year, especially, study of the sources by optical means has 
flourished, and some fascinating hitherto unknown attributes of the universe 
have been revealed. The fascination is all the greater because there are 
suspicions that radio telescopes are receiving signals beyond the range of 
the most powerful optical telescope in the world, the 200-inch Hale telescope 
at Palomar. One of the most exciting astronomical frontiers therefore 
demands a very close collaboration between radio astronomers and optical 
astronomers. The Observatories during the year were in the midst of this 
search, in part because of the capacity of the Hale telescope to see farther 
into space than any other instrument on earth. 

Indications of the highly unusual characteristics of many of the radio 
sources first came in 1960 with R. L. Minkowski's success in obtaining 
positional photographs and spectra of the galaxy 3C295, previously observed 
in radio surveys undertaken at Cambridge, England, and at the Owens 
Valley installation of California Institute of Technology. Minkowski's 
spectra showed 3C295 to be, by a wide margin, the most distant object 
observed up to that time in the universe. The large redshifts of these radio 

1 Letter of August 3, 1964. 

2 Radio, infrared, optical, ultraviolet, and X-ray radiation. 


objects alone would attract astronomers' attention and arouse further 
curiosity because of their possible evidence on cosmological problems. 
Research during the year, however, has shown that even more unusual 
objects are to be found among the known radio sources than had first been 
suspected. Optical study of radio galaxies continues to be a very productive 
source of new data on the universe. The most exciting discoveries concern 
the quasi-stellar sources, very remote objects that probably are highly 
unusual galaxies, having about fifty times the intrinsic brightness of a 
normal galaxy. 

The unique nature of the quasi-stellar sources was first determined with 
certainty last year and reported upon in Year Book 62. In the program 
carried on in cooperation with T. A. Matthews of the Owens Valley Radio 
Astronomy Observatory, six objects of a stellar appearance had previously 
been identified by A. R. Sandage. M. Schmidt's examination last year of 
the spectra of two of them, 3C273 and 3C48, showed unexpectedly large 
redshifts, identifying them as objects far more distant and as much as a 
hundred times brighter than had been thought. Previously, 3C273 had been 
given an accurate position by C. Hazard of Sydney University and J. 
Shimmins of the Commonwealth Scientific and Industrial Research Organ- 
ization, Australia. It turned out to be an object known for at least 80 years 
but assumed to be a normal star. 

It was suspected that these objects were the most powerful radiators of 
energy yet discovered in the universe. Their discovery led to a much- 
increased research effort on the quasi-stellar sources both at the Observa- 
tories and elsewhere during the past year. Sandage, using a technique that 
compares ultraviolet radiation with blue and yellow emission from the 
quasi-stellar source, continued a program of optically identifying known 
radio sources of this type. All quasi-stellar sources have intense ultraviolet 
emission in comparison with other stellar or galactic sources. An ingenious 
technique, which begins with double-image photography using a blue and 
ultraviolet filter on the 100-inch reflector at Mount Wilson or the 48-inch 
schmidt telescope at Palomar and ends with photoelectric observations on 
the Hale telescope, has made absolute identification possible. In this way 
the number of known quasi-stellar sources was more than doubled, being 
thirteen and possibly fourteen at the end of the report year. Matthews' 
parallel search produced seven additional optical identifications of radio 
sources that may be quasi-stellar, raising the total tentatively identified 
to twenty. 

Meanwhile, Schmidt undertook to obtain additional data on redshifts 
from the known quasi-stellar sources. Two of those investigated, 3C47 and 
3C147 (see plate 1), whose radio positions were found by Matthews, showed 
respective redshifts of 0.425 and 0.545, corresponding to velocities of 
recession from the earth of nearly half and more than half the speed of light. 
Thus the source 3C147 now proves to be the most distant object yet found 


in the universe. Yet its brightness is of the 18th apparent magnitude, 3 
indicating a radiation source of stupendous proportions, even in astrono- 
mers' terms. (The faintest objects observable with the Hale telescope are 
of the 23rd magnitude.) 

Although the radio properties of the quasi-stellar sources differ little from 
those of most radio galaxies the optical properties are very different. Indeed, 
they seem to be a distinct new class among the known objects of the sky. 
The brightest of the four quasi-stellar sources that have been studied thus 
far (3C273) has an absolute visual magnitude of —26, making it about 
2.5 trillion times the luminosity of the sun. 

The nature of physical processes within these extremely luminous objects 
is still very puzzling, even though it has commanded the attention of many 
astrophysicists and not a few astronomers of note. According to J. L. 
Greenstein and Schmidt, in the model that most directly corresponds to 
data thus far observed the radio emission comes from a relatively large 
region surrounding a luminous nebula. The radii of 3C273 and 3C48, the 
two nebulae studied carefully, are thought to be, respectively, about 1 
parsec and possibly 2 parsecs (about 1850 billion miles). Within the very 
bright visible nebula that produces the emission lines there is thought to be 
a much smaller, extremely dense object which may be the source of the 
continuous spectrum. In 3C273, this inner core may have a radius one- 
seventieth of the gas nebula and a mass of at least 100 million suns. The 
continuous spectrum of this same object may result from a mixture of 
hydrogen recombination, two-photon emission, and synchrotron radiation. 4 
Although the quasi-stellar objects probably resulted from massive explo- 
sions, their unusual spectral qualities, including the optical thickness of the 
ionized hydrogen regions, raise difficulties for every model of the energy- 
transfer processes within them. When their physical nature has finally been 
resolved it may be found to be very different from anything familiar in the 
more normal objects in the sky. Meanwhile it is certain that the Observa- 
tories have had an important role in placing a tantalizing problem before 
astronomers all over the world. 

The optical identification of galactic radio sources and the identification 
of their properties also produced some very interesting results during the 
year. Among them are data on the classification of different radio galaxies 
and information on the source of cosmic rays. In an investigation of the 
optical and radio properties of forty-two well identified radio sources, 

3 Magnitude is the measure of brightness of a source of radiation in the skies on an accepted 
astronomically determined quantitative scale. Brighter objects are negative on the magnitude 
scale; fainter objects are indicated by positive numbers. The larger the positive number, the 
fainter the object. The brighter stars visible to the naked eye are of the 1st apparent magnitude, 
and the faintest visible (100 times less bright) are of the 6th apparent magnitude. Absolute magni- 
tude is the apparent magnitude that an object would have if viewed at a standard distance of 
32.6 light years. 

4 The light emitted by an electron or proton when accelerated in a magnetic field. 


Matthews of the California Institute of Technology found a strong simi- 
larity among most of them. He found that most of the sources are spheroidal 
galaxies, with luminous nuclear regions of elliptical shape, surrounded by 
extensive envelopes of varying visibility. Such a galaxy is often the brightest 
and largest in a cluster of galaxies, although not all radio galaxies occur 
within clusters. All the weaker radio sources are spiral or irregular galaxies. 
They are sharply differentiated in their emission from the strong radio 
sources, which are the spheroidal galaxies and quasi-stellar objects. 

The indication of cosmic-ray sources came from Sandage's continued 
study of the nearby radio galaxy M 82, which he and Lynds of the Kitt 
Peak Observatory had started in 1963. (See plate 2.) Studying the extensive 
outer system of blue filaments that stretches as much as 130,000 light years 
above and below the fundamental plane of the galaxy, Sandage and Miller 
have obtained direct evidence of an organized large-scale magnetic field 
extending along the minor axis of the galaxy. Light from parts of the blue 
filaments was found to be about 80 per cent polarized. Assuming the 
magnetic strength of the fields to be as calculated by Sandage and Lynds 
last year, Sandage states that the electrons must have extremely high 
energies if they are to be compatible with the existence of synchrotron 
radiation and optical frequencies. The energies he postulates are in the 
range of cosmic-ray energies. He considers the year's observations as 
providing further evidence that the origin of cosmic rays throughout the 
universe is connected with the violent energy-transfer processes in radio 
galaxies and other radio sources. Thus one of the longest standing mysteries 
in the world of physics and astronomy, the sources of the cosmic rays that 
have been known for several decades, moved a step closer to solution. 

Observations during the year also added new knowledge for the interpre- 
tation of the early history of our own galaxy and a revised conception of 
its shape and classification. Sandage and B. N. Katem completed a four- 
year program of three-color photoelectric photometry of five star clusters, 
M 3, M 13, M 15, M 92, and 47 Tucanae. Analysis of these globular clusters 
showed that their " horizontal branches" 5 have the same visual magnitude 
regardless of metal abundance. The data accordingly suggest that the ages 
of the clusters in the galactic halo fall within a range of ±20 per cent. 
Adopting the evolutionary model calculated by F. Hoyle of Cambridge 
University, these observations confirm the report made last year giving a 
new age determination of approximately 12 billion years for the galaxy. 

H. C. Arp, after extensive observations in the direction of the center of 
the galaxy, concluded that most of the stars near the galactic center are 
older than a billion years. His data also demonstrated that the nucleus of 
the galaxy is only about half the size of the nucleus of the Andromeda 

5 A diagram of stars plotting temperature (color) -magnitude relations shows a "main sequence' 
that includes most stars. "Branches" are deviations from the main sequence on the diagram. 


nebula. Our galaxy, therefore, should be classified as an Sc rather than an 
Sb system, as it has been until now. According to Hubble's system our 
galaxy is a "late spiral" rather than an "intermediate" system. The Sc 
spiral is generally more open than the spiral of the Sb type. (See plate 3.) 

During the year the Observatories also took steps that eventually may 
lead to exploration of the southern skies with the same thoroughness that 
the Hale telescope and their other instruments have made possible for the 
northern skies. With the cooperation of the Associated Universities for 
Research in Astronomy and the University of Chile, the Institution set up 
two automatic seeing monitors on favorable mountain sites in central Chile, 
after reconnaissance by Babcock. Plans were also laid for seeing monitor 
observations in south central and southwestern Australia as well as on 
Palomar Mountain. These observations are expected to provide more 
precise data than those hitherto available, permitting quantitative com- 
parison of astronomical seeing conditions for three continents. 

The pressing need felt by astronomers for good data on the southern 
skies was illustrated by the Observatories' efforts during the year to extend 
the National Geographic Society-Palomar Observatory Sky Survey on the 
48-inch schmidt telescope into the declination zones —36° and —42°. With 
an improved emulsion and filter combination, plates were obtained by 
J. B. Whiteoak that are considered temporarily usable until adequate 
equipment can be placed in the southern hemisphere. The Institution hopes 
that eventually funds may be obtainable to establish an improved version 
of the 200-inch telescope in the southern hemisphere, along with needed 
auxiliary instruments like a 48-inch schmidt telescope and a 60-inch photo- 
metric telescope. 

Whiteoak's southern sky survey, and the preliminary investigation for a 
major southern hemisphere observatory (CARSO), typify the spirit in 
which the combined observatories and their predecessor, the Mount Wilson 
Observatory, have been managed for nearly fifty years. Even though 
astronomical research has been, and still is, very much an individual matter, 
it has always been heavily dependent on unique instruments and rare 
techniques. They include not only the great telescopes but also the capacity 
to produce diffraction gratings for spectroscopy and the exceedingly exacting 
demands for photography. As the world's principal repository of these 
instruments and techniques, for many years the Observatories have been 
managed with a sense of responsibility toward all astronomers, not the 
Observatories' staff alone. Thus the production of gratings, which began in 
1912 and was finally terminated in 1964, resulted in the distribution of 25 
gratings to other observatories scattered over the world. 

Typically, also, observing time on the Palomar and Mount Wilson 
instruments has a large component for guest observers. During 1963-1964 
there were twenty-five guest observers : six from foreign observatories, five 


from other American observatories, nine from American universities and 
colleges, four from American private corporations, and one from the United 
States federal government. The Institution's hopes for and preliminary 
work toward a CARSO 200-inch southern hemisphere telescope is a contin- 
uation of this policy and tradition. Although other organizations have plans 
for southern hemisphere observatories, the proposed CARSO 200-inch is 
the only one of its size for which complete plans have been developed thus 
far. Mindful of the unique contributions of the Hale telescope to astronomy, 
the Observatories consider a southern hemisphere 200-inch telescope 
absolutely essential to the immediate future of astronomy. 

The Committee on Image Tubes for Telescopes 

The same spirit has prompted the Institution's participation in, and the 
leadership it has provided for, the Committee on Image Tubes for Tele- 
scopes, which marked its tenth anniversary in 1964. This Committee, 
chaired by M. A. Tuve, Director of the Department of Terrestrial Mag- 
netism, is in some ways at the opposite extreme from the typical research 
project in astronomy. It is not only a group undertaking; it is interdepart- 
mental within the Institution; it is interorganizational ; and its work has 
been carried forward by nonprofit organizations, federal government 
agencies (National Bureau of Standards and United States Naval Observa- 
tory), and private industrial corporations. Besides Tuve, the present 
Committee members are W. A. Baum of the Mount Wilson and Palomar 
Observatories, J. S. Hall of the Lowell Observatory, and L. L. Marton of 
the National Bureau of Standards. 

From its inception the Committee had as an objective development of 
methods of using the high quantum efficiency of photoemissive surfaces for 
astronomical observations. It was realized that relatively high gains in 
sensitivity might be obtained by means of such devices, if they could be 
developed, that would greatly increase the effectiveness of even moderate- 
sized research telescopes. Although the Committee is not alone in attempt- 
ing to develop image tubes, 6 its ten-year program, supported by the National 
Science Foundation, appears to be the most intensive and methodical one 
directed toward this objective. It is a source of satisfaction to report that 
during the year the Radio Corporation of America manufactured experi- 
mental tubes for the Committee that for the first time meet the very 
stringent requirements for use in telescopes. 

Two of these tubes, called cascaded converters, were tested at the 
Department of Terrestrial Magnetism laboratories, at the Lowell Observa- 

6 Lallemand of France had a pioneering program of image converter development, and McGee 
of Great Britain is working on a "Lenard-window" type of tube. The Soviet Union also has an 
image tube development program. 


tory, and at Mount Wilson Observatory. The tubes were compared with 
the normal methods of collecting information 7 by telescope. Tests by 
W. K. Ford at the Department of Terrestrial Magnetism laboratory showed 
a rate of information gain of 30 as compared with the best performance of 
normal telescope photography. Spectroscopic tests at Lowell Observatory 
showed a speed gain for equal resolution of about 10 for the image tube 
system. Tests by Ford and Baum on the Mount Wilson 100-inch telescope 
showed a rate of blackening for spectroscopic plates eight times higher than 
that of unaided photography. On the basis of these tests the Committee 
concludes that a gain of about 10 in the rate of collection of information 
may be obtained at a telescope on which the tubes are used. The Committee 
therefore has placed an order with the Radio Corporation of America for 
twenty of these high-performance tubes. This order has been made possible 
by financial support from the National Science Foundation. When manu- 
facture has been completed the tubes will be distributed for use at various 
observatories under the guidance of the Committee and the National 
Science Foundation. 

The Department of Terrestrial Magnetism 

It was noted above that the Department of Terrestrial Magnetism has 
not merely participated but has taken a leading role in the development of 
image tubes. A pattern of group effort and interorganizational cooperation 
is a familiar one to that Department. As the reports in this Year Book 
illustrate most strikingly, the year's studies in seismology, electrical con- 
ductivity anomalies, isotope geology, and biophysics show an interesting 
pattern of cooperating group effort. Here the "explorers" seem to form in 
search parties. Although these groups are nominally responsible to a section 
chairman and also of course to the departmental director, their work can 
best be described as genuine collaboration among peers. In general it is 
association by choice, since Institution custom recognizes the right of any 
Staff Member to work in another manner if he so wishes. 

Geomagnetic and Seismic Studies 

For a number of years the Department of Terrestrial Magnetism has been 
interested in developing South American talent in geophysics. Interest has 
centered particularly on the problems of the Andean area, a key region in 
research on many geophysical phenomena. The first interests of the Insti- 
tution go back to the Huancayo magnetic observatory, started in 1922, and 
donated by the Institution to the Government of Peru in 1947. More 

7 "Information" is considered to include image resolution, "noise" or granularity on photo- 
graphic plates, and the rate of blackening. 


recently cooperation in seismic studies was begun. Both types of study 
continued during 1963-1964, in the most intensive effort yet marshaled. 

Since 1962 a cooperative study of geomagnetic anomalies in central and 
southern Peru has been conducted by S. E. Forbush, U. Schmucker, and 
0. Hartmann of the Institution; A. A. Giesecke and M. Casaverde of the 
Instituto Geofisico del Peru, with the additional cooperation of the Uni- 
versidad Nacional de San Agustin, Arequipa, Peru. Financial support from 
the National Science Foundation to the Instituto Geofisico del Peru, and 
instrumental loans from the United States Coast and Geodetic Survey, 
assisted the project. To investigate suspected anomalies, nine temporary 
observing stations were set up between Huancayo, in central Peru, and 
Ayanquera, about 420 miles farther south. Observations from these stations 
during the year, particularly during times of magnetic storms like that of 
September 21, 1963, clearly show a remarkably high electrical conductivity 
at shallow depths beneath parts of the Andes Mountains in southern Peru. 
(See fig. 1.) 

Electrical currents are induced in the earth's crust and mantle by time- 
varying electrical currents in and above the ionosphere. The depths at 
which the induced earth currents generally flow has been estimated at about 
150 to 250 miles, indicating that the temperature of rock materials there is 
high enough so that they are good electrical conductors. Rock materials at 
a temperature of 1500°C have, according to experiment, an electrical 
conductivity about 10,000 times greater than such materials at 500°C. The 
Department's studies of the seismic properties of the earth's crust under 
the Andes, and studies of the equatorial electro jet in the ionosphere over 
Peru, indicated that strong induced earth currents were to be expected at 
relatively shallow depths under the Andes. 

Evidence confirming this hypothesis was obtained during 1962 and 1963 
by the Department and the Instituto Geofisico del Peru in a limited survey 
of geomagnetic variations simultaneously recorded at nine stations 100 or 
200 kilometers (60 to 125 miles) apart in southern Peru. A preliminary 
study of these observations showed anomalous induced earth currents in 
the region near Cuzco, probably at a depth of 20 miles or less. Near the 
coast of southern Peru, the induced earth currents at similar depths were 
evidently sufficient to mask the normal coastal anomaly caused by the high 
conductivity of the ocean. 

One reasonable interpretation of these unusual high conductivities found 
at shallow depths is that rock materials of abnormally high temperatures 
for their depth may underlie parts of southern Peru, as they also are thought 
to underlie parts of Germany and Japan. Thus, Forbush, Schmucker, 
Hartmann, Giesecke, Casaverde, and their associates may have found 
another "hot spot" near the earth's surface. It will be interesting to see 
whether this explanation can be corroborated by other geophysical evidence. 



Geomagnetic stations o 
Seismic stations + 

_ , O 100 200 300 400 

Scale of miles i i l_ 

Fig. 1. Location of Carnegie Institution cooperative geomagnetic and seismic stations in South 



Seismic studies in the Andes during the year produced another interesting 
general observation about the earth's interior. The Department maintained 
a collaborative study with a half dozen university geophysics groups in 
Peru, Bolivia, Chile, and Argentina. Members of this group, composed of 
M. A. Tuve, I. S. Sacks, L. T. Aldrich, J. S. Steinhart, and R. Sumner of 
the Institution; J. Frez, Fr. G. Saa, E. Gajardo, and C. Lomnitz of the 
Universidad de Chile; R. Anzoleaga, J. Santa Cruz, and R. Salgueiro of the 
Instituto Geofisico Boliviano; R. Cabre of the Observatorio San Calixto, 
Bolivia; F. Volponi of the Universidad de Cuyo, San Juan, Argentina; and 
A. Rodriguez of the Universidad de San Agustin, Arequipa, Peru, have 
maintained seismographs providing comparable records at about thirty 
different sites in the Andes. Coordination of study methods and seismograph 
location, and local study according to a coordinated plan, promise to pro- 
vide very useful data on the great irregularities of structure to be found in 
the crust and upper mantle of this tectonically active region. The repeated 
occurrence of large and small earthquakes many times daily gives a profusion 
of materials for study which makes this area an especially attractive avenue 
for fresh information about the earth's interior by seismic activity. 

One of the studies growing out of this group effort during the year led to 
a revised and more accurate measure of the diameter of the dense interior 
core of the earth. A study of the seismic compressional waves diffracted 
around the boundary of the earth's core showed that the shadow boundary 8 
is about nine degrees less than estimated in calculations accepted hitherto. 
The radius of the core at the points from which the studied compression 
waves were diffracted thus appears to be about 2 per cent larger than in 
previous estimates. The new figure would appear to be about 2120 miles. 

Such a revision of the core dimensions has several significant sequels for 
geophysics. In particular it requires a substantial revision of the velocities 
at which seismic waves are estimated to travel in the lower mantle. It also 
raises the question whether the boundary of the core has a uniform curva- 
ture. The seismic studies group has begun an investigation of the range of 
curvatures of this boundary, and has designed for the purpose an improved 
seismometer that offers more accurate measurement of long-period vibra- 
tions than older instruments, is temperature insensitive, and can be operated 
in remote areas. 

The Isotope Geology Group 

Another interesting research group, in which Staff Members of both the 
Department of Terrestrial Magnetism and the Geophysical Laboratory 
participate, is the isotope geology group. More tightly organized than the 

8 Compressional waves travel through the mantle at a higher velocity than through the core. 
Waves entering the core thus are "diffracted." The boundary between waves traveling directly 
through the mantle, but just grazing the core, and diffracted compressional waves is the shadow 


seismic studies group, its members have been particularly interested in the 
history of the crust and outer mantle of the earth, on which they have 
provided an abundance of new data within recent years. " Charter" mem- 
bers of this group are L. T. Aldrich of the Department of Terrestrial 
Magnetism, and G. R. Tilton and G. L. Davis of the Geophysical Labora- 
tory, who started work together in 1951; they were joined later by S. R. 
Hart of the Department of Terrestrial Magnetism. During this year Rama, 
of the Tata Institute, Bombay; J. Richards, of the Australian National 
University; J. Gerken, of the Universidade de Minas Gerais, Brazil; R. H. 
Steiger, of the Swiss Federal Institute of Technology, Zurich; and P. W. 
Gast, of the University of Minnesota, were also members of the group. 

Although isotope dating of crystalline and metamorphic rocks is now 
undertaken at a number of universities and other research institutions both 
in the United States and abroad, 9 the Carnegie Institution group has been 
particularly influential in pioneering the application of comparative dating 
techniques, using as many decay systems as possible (rubidium/strontium, 
potassium/argon, and uranium/lead, especially). As part of a study of the 
ultramafic 10 rocks of the oceanic basins the group completed this past year 
a detailed study of the potassium, rubidium, and strontium geochemistry 
of St. Peter and St. Paul Rocks. These islets lie on the mid-Atlantic ridge, 
and have a unique association of rocks that has led some geologists to 
suggest that they are surface exposures of the oceanic mantle. They are 
considered very significant for testing models of ocean basin evolution. 

All the samples taken were a mylonitized peridotite, a rock composed of 
a relatively high proportion of iron- and magnesium-containing minerals 
that has been subject to breakage and deformation under great pressure. 
Peridotite is thought to be a major component of the outer mantle. In the 
course of analyzing the samples the interesting question arose whether one 
or more was not the actual material of the original unaltered mantle of the 
earth. An important indicator of this possibility is the strontiums/stron- 
tium 86 ratio of the samples. 

One of the St. Peter and St. Paul Rocks samples has a ratio clearly 
higher than that of the basalts that are so characteristic of the ocean basins. 
The isotope group accordingly states that it is not a product of the normal 
oceanic rock generation process. On the other hand, if the mantle was 
formed 4.55 billion years ago containing strontium of a Sr 87 /Sr 86 composition 
slightly higher than that detected in achondritic meteorites, 11 four of the 

9 Notably at Massachusetts Institute of Technology, Lamont Geological Observatory of 
Columbia University, United States Geological Survey, California Institute of Technology, 
University of California at Los Angeles, and University of California at Berkeley, in the United 
States. Isotope research groups are found abroad in Australia, England, South Africa, Switzerland, 
Finland, France, Italy, Japan, and the Soviet Union. 

10 Mafic is a mnemonic for minerals containing a high ferromagnesium content, in contrast to 
the felsic minerals, containing higher contents of potassium, sodium, calcium, and aluminum. 

11 Achondritic meteorites are meteorites that lack chondrites, spheroidal masses of varying 
mineral composition found in the groundmass of chondritic meteorites. 


samples could be interpreted as having a 4.5-billion-year age. The group 
concludes that these four samples are consistent with identification as 
chemically unaltered primordial mantle, and one sample in particular 
cannot easily be explained in any other way. It is fascinating to think that 
here at last we may have a physical relict of the composition of the outer 
shell of the earth at so very early a date in its history. 

At the same time another investigation by the group has suggested that 
a revision of the age of the earth, as assumed in the above paragraph, may 
be in order. The accepted value of about 4.55 billion years was obtained by 
comparing the isotopic composition of "modern terrestrial lead" with that 
of lead from the troilite (ferrous sulfide) phase of iron meteorites, postulated 
to be the same as primordial lead. However, an analysis of the isotopic 
composition of lead in feldspars and galenas of different ages on the North 
American continent suggests that the earth may be somewhat older. The 
4.55-billion-year age was calculated by assuming that modern lead had 
evolved over the eons in a chemically closed system. Instead, the new North 
American data indicate that the lead sources have been enriched with 
uranium by chemical transport and hence were not in a closed system. The 
inapplicability of the completely closed system suggests a minimum age of 
4.7 billion years. Since the area from which the samples were taken was a 
rather limited one, it will be interesting to see whether the provisional new 
age is confirmed by isotopic studies from other areas of the earth. 

Finally, the group used isotopic analysis to test an hypothesis about the 
origin of crystalline and metamorphic rock masses in continental areas. 
The isotopic composition of lead and strontium in surface rocks in terres- 
trial areas is known to be highly variable, depending on rock type and age. 
The hypothesis was that lead in rocks of deep-seated origin like basalt 12 
might have a narrowly defined isotopic composition. Accordingly, it could 
be used as a tracer to identify rocks of deep origin and distinguish them 
from others containing admixtures of surface materials. If it could be shown 
that volcanic rocks from oceanic areas, which are relatively uncontaminated 
by lead and strontium from continental areas, have a relatively homo- 
geneous isotopic composition, some confirmation of the hypothesis would 
be obtained. 

With this in mind the isotopic composition of lead in volcanic rocks from 
Gough and Ascension Islands was analyzed. It was found that the isotopic 
composition of lead in basaltic rocks was not uniform on the two islands, 
and that the lead from basalt was less radiogenic than that from trachyte 
(a more siliceous rock than basalt) on either island. Data for the isotopic 
composition of strontium in the Gough and Ascension samples showed 
similar deviations. Therefore, it was concluded that isotopic composition of 
either lead or strontium does not always offer a convenient tracer to deter- 

12 Basalt is a finely textured igneous rock exposed at the earth's surface in lava flows. It is 
typically composed of feldspar and pyroxene (ferromagnesian) minerals. 


mine the origin of surface rocks. Rocks of deep origin apparently have 
radiogenic differences as well as the rocks formed partly of crustal materials. 

The Experimental Petrology Group, the Geophysical Laboratory 

Another vigorous group devoted to geophysical studies is that carrying 
out experimental petrology investigations at the Geophysical Laboratory. 
Its broad objective is a description of the earth's interior, particularly the 
mantle portion. It may be worth lingering a moment on a description of 
the structure and working methods of this group, because they are typical 
of the operations of the Geophysical Laboratory, and they illustrate work 
habits among other groups in the Institution. 

The group is informally organized, and varies both in size and composition 
from year to year. Senior Staff Members with continuing interest in 
petrology or petrography, like F. R. Boyd, F. Chayes, J. L. England, 
D. H. Lindsley, J. F. Schairer, and H. S. Yoder, Jr., have composed a core 
group for some years. Usually one or more senior guest investigators from 
other institutions join the group on invitation. This year C. E. Tilley of 
Cambridge University and A. E. Ringwood of the Australian National 
University were in residence at the Laboratory. B. J. Skinner of the United 
States Geological Survey also participated in some of the group's work. 
Finally, a group of skilled younger men, desiring the stimulus of the group 
environment, participate as Fellows. During this year they were D. K. 
Bailey, Trinity College, Dublin; P. Bell, Harvard University; B. T. C. 
Davis, Princeton University; J. J. Fawcett, University of Manchester; 
I. Kushiro, Tokyo University; I. MacGregor, Princeton University; 
D. Metais, University of Paris; D. Presnall, Pennsylvania State University; 
and B. Velde, Montana State University. 

The group continues a tradition of investigating relations in mineral 
systems pioneered by the Laboratory almost sixty years ago. Studies in 
this same field have also been carried out now for a number of years at other 
institutions in this country and abroad, 13 but the experimental petrology 
group continues to occupy a leading place in these studies. Such a place has 
been assured by the group's continuing development of experimental 
techniques and equipment. Creation of very high pressure equipment for 
synthetic mineral formation has made possible the investigation of some 
fundamental questions about the nature of the earth's interior. Although 
there is agreement on the priority of these questions, senior investigators 

13 Within the United States, experimentation on mineral systems at high pressures is under- 
taken at Harvard University; Pennsylvania State University; University of Texas; University 
of California at Los Angeles; University of Chicago; United States Geological Survey; University 
of Illinois; Air Force, Cambridge; Battelle Memorial Institute; E. I. du Pont de Nemours & 
Company; General Electric Company; and Norton Company. Leading foreign centers of study 
are in the Soviet Union, Japan, Great Britain, and Australia. 


make their own choices about pertinent experiments or other studies. The 
ideal seems to be, as Professor Robert Bowie 14 of Harvard has expressed it 
in another context, an environment that stimulates the capacities and 
broadens the scope of individual research workers, achieving a degree of 
coherence without cramping individual work. Through the simple mech- 
anism of residence under one roof, and other devices such as seminars and 
the Penologists' Club, the group finds opportunities to evaluate hypotheses 
and other ideas, criticize methods, and examine results. The productiveness 
of these methods is proved not only by the results reported in this and past 
years but also by the attractiveness of the Institution fellowships to 
promising young geologists throughout the world. 

The 1963-1964 year saw the experimental petrology group emerge with 
some significant clarification on the mainstream problems of petrology to 
which many years of effort have been devoted. The many hundreds of 
known minerals and their even more numerous assemblages in rocks have 
presented a bewildering array to anyone interested in petrologic evolution. 
Classification was first approached in modern times as a problem of chemical 
and physical composition. However, ordering minerals on this basis gave 
little information on the genetic relations among the minerals and the 
igneous rocks that are the primary materials of the earth's mantle and crust. 
Seeing this lack, the experimental petrology group for some years has held 
the working hypotheses that (a) the many important igneous rock types 
observable in the world are not just random accumulations of minerals, and 
(6) the basalts are keys to the evolution of most igneous rocks. Basalts are 
considered important not only because they form a large part of the earth's 
crust but also because the liquids that crystallize as basalt originate deep 
within the earth (below 40 kilometers) and therefore are evidence of the 
processes occurring there. Indeed, they give us partial chemical samples of 
what may be found in the earth's interior. (See plate 4.) 

Four years ago ( Year Book 59) the first reports were given on a model of 
phase relations that suggest the evolutionary relationships existing among 
important basalt types. (See fig. 2.) This work has since been extended and 
revised. It is gratifying to report this year that as a result we now have 
another window on the wonderfully ordered structure of nature. For the 
first time it is possible to outline the overall design of the fractionation of 
major magmatic types that are the parents of a large grouping of basalts. 
It is possible to express these relations, which cover a group of rock types 
with a vast array of mineral compositions, with one simple flow sheet 
(fig. 3). It is even possible that a single, although heterogeneous, magmatic 
source exists for them. 

All the experiments from which the flow sheet was derived were conducted 

14 Robert R. Bowie, Newsletter, the Harvard Foundation for Advanced Study and Research, 
May 15, 1964, p. 1. 



Fig. 2. Diagrammatic representation of phase relations among minerals composing alkali basalts. 
Directions along the tetrahedron show increasing content of given sets of elements. For example, 
increasing silica content is shown in the directions Ne-Qz, Fo-Qz, and La-Qz; increasing magnesium 
content in the lines converging on Fo, calcium in those converging on La, etc. The tetrahedron 
describes relations in basalt mineral formation at 1 atmosphere pressure; temperatures of forma- 
tion differ. 

Ab, albite, NaAlSi 3 8 
Ak, akermanite, Ca 2 MgSi 2 0i 
Di, diopside, CaMgSi 2 Os 
En, enstatite, MgSi0 2 
Fo, forsterite, Mg 2 Si04 
La, larnite, Ca 2 Si04 

Mer, merwinite, Ca 3 MgSi 2 07 
Mo, monticellite, CaMgSi0 4 
Ne, nepheline, NaAlSi0 4 
Ra, rankinite, CasSi 2 07 
Sm, soda melilite, NaCaAlSi 2 07 
Wo, wollastonite, CaSi03 

at normal atmospheric pressure (1 atmosphere). The experiments sum- 
marized in the flow sheet showed that the major magmatic types, often 
closely related in their field occurrence, have "thermal barriers." However, 
the separation of pyroxene minerals from the magma at different pressures 
(i.e., different depths within the earth) is considered a promising possible 
explanation of these discontinuities in the flow diagram. Studies of some of 




Mg 2 Si0 4 


Mg 2 Si0 4 


CaMgSijOj Lherzolite^ A 
SiO, * 







Mg 2 Si0 4 

CaMgSi 2 6 



Basanite Olivine 



Wollastonite -nepheline 
Tephrite Wollastonite 


CaSi0 3 

CaMgSi 2 4 


NaAISi0 4 

NaAISi 3 8 


Co 2 MgSI,0 7 
CoM g 'Si 2 4 



Olivine -melilite NaAisio 4 
Nephelinite c a2 MgSi 2 o 7 

" CaMgSi 2 4 


Wollastonite -melilite 
Nephelinites CaSi ° 

Ca 2 MgSi 2 7 
CaMgSi 2 4 

NaAISi0 4 

Ca,MgSi 2 7 


Fig. 3. Simplified flow sheet, showing relations of major basalt types extruded at the earth's 
surface. Each has a characteristic set of minerals. Arrows indicate directions of transition; the 
"watersheds" between arrows of different direction are temperature barriers, not crossed at 1 
atmosphere pressure by magmas of given composition; and the lettered positions A, B, etc., are 
points of equilibrium. The names are those of the principal basalt types, usually reflecting a 
characteristic mineral component. The names of the component minerals whose symbols appear 
on this figure are given in the legend of figure 2. 

the minerals characteristic of the alkali basalts, especially indicative of 
pressure relationships, showed that pressure or depth control is indeed 
important in determining the particular mineral components of a given 
rock type. The performance of pyroxene solutions at high pressures has led 
to the interesting speculation that a single magma could yield the two 
principal types (tholeiite and alkaline) into which all the basalts of the 
world are classified. Experiments during the year examined this possibility 
and suggested that there are indeed discernible relations between these 
principal basalt types. Although these experiments only suggest possibilities 
at this point, they do not preclude a single source magma at depth within 
the earth. 

The group's experiments during the year also provided confirming evi- 
dence for the validity of a model of the earth's upper mantle. Geothermal, 
seismologic, and petrologic evidence has led to a postulate that the dominant 


rock in the upper mantle is an aluminous peridotite. 15 Because of observed 
earth temperature gradients and the natural distribution of peridotite types 
that have been erupted from the mantle to the surface, the model also 
suggests a type of peridotite known as spinel 16 peridotite to be present 
beneath the oceans to a depth of about 60 kilometers but absent or very 
thin beneath the great crystalline "shields" like that of Canada. Finally, 
seismic data indicate a discontinuity at a depth between 150 and 250 
kilometers suggesting a gradual change in rock type at that depth. Another 
part of the model postulates the primary composition of the upper mantle 
as lying between that of basalt and of peridotite. 

The experiments of Boyd, Ringwood, and MacGregor, during the year, 
showed that a peridotite composed of about 1 part basalt and 3 parts 
dunite 17 will crystallize into three different types, each stable under different 
pressure-temperature conditions. The three types, in the order of increasing 
pressure and temperature of formation, are spinel pyrolite, pyroxene 
pyrolite, and garnet pyrolite. The observed laboratory pressure-temperature 
conditions under which the three types are stable fit the model of the upper 
mantle determined by seismic and earth temperature observations. Of 
particular interest is the fit of these data to the zone of seismic discontinuity 
revealed by the change in velocity of the compressional waves between 150 
and 200 kilometers. The experiments indicate that this is a zone of transition 
from pyroxene pyrolite to garnet pyrolite. Furthermore, laboratory- 
determined estimates of seismic velocities in these two rock types correspond 
to the observed change in the earth itself measured by seismologists. Thus 
a firmly based concept of the structure of the earth deep within the mantle 
appears to be emerging, as well as a picture of the surprising dynamic 
continuity of its processes and those that form the crustal zone. 

Another study by the petrology group deserves mention as an illustration 
of an unorthodox approach to a problem, and the potential results from it. 
Chayes has been studying the possibilities of applying discriminant func- 
tions to petrologic problems. The basic techniques were developed in the 
1920s and 1930s by Fisher, Pearson, Mahalanobis, and other mathemati- 
cians and statisticians for use in the life and behavioral sciences. However, 
these methods did not come into their own until means of high-speed 
computation became generally available. Among the most useful of these 
statistical techniques is the discriminant function. Interest in it was renewed 
in the late 1950s, and Chayes has recently been a leader in applying it to 
geologic problems. 

During the year Chayes and his collaborator, D. Metais, continued the 
study of distinctions between basalts of the oceanic islands and those on 

15 A coarse-grained rock containing a magnesium silicate (olivine) and pyroxene minerals. 

16 Spinel is typically magnesium aluminate, but it may include iron and a few other elements. 

17 A peridotite composed principally of olivine and chromite. 


the shoreward margins of the open oceans. These two general types of basalt 
can be distinguished by the higher titania (Ti0 2 ) content of the oceanic 
basalts. By this means, as noted in Year Book 62, they classified large 
assemblages of samples into the two types with an efficiency of more than 
93 per cent. Discriminant function analysis now shows that no weighted 
linear combination of components not including titania is effective in 
classifying bulk samples into the appropriate group. They conclude that 
discriminant function analysis will materially facilitate petrographic 
classification. However, Chayes adds sagely in his report, "discriminant 
function analysis is a supplement to but not a substitute for contemplation." 

The Biophysics Section 

One of the most tightly knit research groups in the Institution is the 
Biophysics Section of the Department of Terrestrial Magnetism. This 
group, formed in the Department shortly after the end of the second world 
war by physicists who wished to apply their skills to the study of biological 
problems, has almost from the beginning been an interesting illustration of 
the dimensions of "critical mass" in scientific research and the benefits it 
confers. Four of the members, R. B. Roberts, E. T. Bolton, R. J. Britten, 
and D. B. Cowie, have worked together since the early 1950s. Roberts and 
Cowie, along with P. H. Abelson, now Director of the Geophysical Labora- 
tory, were members of the group when it was first formed after the war. 
B. J. McCarthy, formerly of Oxford University, joined the Section more 
recently as a Senior Staff Member, and for the last few years B. Hoyer, of 
the National Institutes of Health, has participated as a visiting investigator. 
D. I. Axelrod, of the National Institutes of Health, also collaborated with 
the Section during the year, and Y. Kato, of the Osaka Municipal Hygiene 
Laboratory, Japan, and M. Miranda, of the Universidade do Brasil, par- 
ticipated in the group's research as Carnegie Institution Fellows during 
the year. 

A major interest of the Section for a number of years after the group's 
inception was the dynamic structure of cells, including the mechanisms for 
nucleic acid and protein synthesis. Their special materials for study were 
Escherichia coli and other bacterial cells as analyzed by physical methods. 
More recently the interests of these men have evolved into a much broader 
range of questions which can be examined by the study of nucleic acid 
interactions. Particularly since the development of the DNA-agar column 
method of analysis two years ago" by members of the group, a near explosion 
of possibilities for using their techniques has taken place. The fascinating 
penetration of these techniques into fundamental biological problems has 
never been better illustrated than during this year, when experiments gave 
new insights on evolutionary relations among animals, gave new insights on 


the manner in which cancer-implicated viruses survive and infect organisms, 
and made a significant contribution to the subject of cell differentiation, 
the process through which a complex animal is developed from the single 
fertilized cell. 

Such a range of subjects places the group in a setting of the whole vast 
sweep of biological research effort that is going forward throughout the 
world. Scarcely a major university in the United States or abroad is without 
some active interest in these problems. In addition, many governmentally 
supported research institutions exist, like the National Institutes of Health 
of the United States, as well as large privately supported organizations, like 
the Rockefeller Institute and the Sloan-Kettering Institute for Cancer 
Research. That this small group has made an increasingly telling impact 
upon so heavily populated and complex a research world not only is a 
tribute to their skills but also, perhaps, is an illustration of what the group 
describes in their report as "the privilege we enjoy to range broadly and 
freely in our ideas and our work." 

The versatility and the power of the DNA-agar technique were again 
demonstrated during the year's research. In a series of experiments investi- 
gating the homologies among the DNA 18 materials of a wide range of species 
in the animal kingdom, the group gave further proof of the genetic related- 
ness among the primates and other vertebrates, and has even been able to 
speculate on the meaning of the homologies for the time scale of evolution. 

Nucleic acids, as the essential materials of heredity, are the root sub- 
stances of systematics and evolution, the group states in this Year Book. 
Using DNA materials from a number of species, including armadillo, calf, 
chicken, hamster, mouse, rabbit, salmon, vervet (a species of monkey), and 
human nucleic acids, the group designed experiments in which the homolo- 
gous DNA of two species was first selected, then tested for homology 
against the DNA of a third species. In this way the degree of correspondence 
within the DNAs of all three species could be discovered. Through these 
experiments, and those of another type called competition experiments, a 
tentative hierarchy of DNAs within the animal kingdom can be constructed. 
When the degree of homology of the DNAs of the different species is com- 
pared with the time at which the organisms are thought to have diverged 
from one another during their evolutionary history, a striking systematic 
relation is revealed. (See fig. 4.) 

The data from the experiments show that the relation is not a random 
process of change. The process of evolutionary change appears to be step- 
like, for when several distantly related species of mammals are examined it 
is found that the homologous DNA fraction any two hold in common is 
also common to all. Furthermore, this mammalian common fraction 
incorporates the smaller fractions that are common to birds and fish. The 

13 Deoxyribonucleic acid. 



Man, chimpanzee 
New world monkey 



100 200 300 400 500 600 
Time since divergence, million years 

Fig. 4. Relation between DNA polynucleo- 
tide similarity (homology) and time of phylo- 
genetic divergence over a period of 600 million 

research workers of the Section therefore suggest that during the course of 
vertebrate evolution about 100 million years elapsed before one-half of a 
DNA segment changed enough in its polynucleotide sequence so that it no 
longer was capable of combining with its evolutionary predecessor, as 
indicated in the DNA-agar procedure. We may therefore infer that an 
important process in vertebrate evolution over eons has been the alteration 
of relatively large segments of chromosomes. This process may have an 
evolutionary significance equal to or possibly greater than that of the 
"point" mutations whose importance has long been recognized. 

Many interesting questions emerge from this glimpse of the DNA 
foundation of evolution. The most important, of course, is the nature of the 
changes that result in the development of a new genome. Interesting 
secondary questions concern the nature of lethal mutations and the process 
that permits the continuance of a species, like the horseshoe crab, unchanged 
for as much as 250 million years. For example, is the 20 per cent of the 
genome common to all mammals absolutely essential to the viability of 
each? Logic would indicate that deletions of genetic material taking place 


in this part of the genome would be lethal. The Section believes that the 
DNA-agar column method is capable of giving us further insights into the 
nature of the mutational changes in the genetic materials. 

The findings of the group that can be related to the etiology of cancer are 
highly specific but nonetheless broadly suggestive for further investigation 
of one important aspect of cancer problems. The first step was the investi- 
gation of the homology relation between a bacteriophage and its host. 

It has long been known that a bacteriophage or virus named lambda 
causes a prompt destruction of Escherichia coli cells in susceptible strains 
so infected. Escherichia coli, of course, is the familiar colon bacillus. How- 
ever, one strain of Escherichia coli known as K12 has two types, distin- 
guished by their responses to ultraviolet light exposure. The normal form 
of K12 shows no effect when exposed to certain specific dosages of ultraviolet 
light; but the other form, called K12 lambda, will abruptly release the 
lambda virus with prompt destruction of most of the cells upon exposure to 
similar dosages. A few survivors from this event may be subcultured as 
often as 50 times with no further sign of phage infection. A subsequent 
ultraviolet light exposure, however, will again promptly release latent virus. 
Clearly the genetic message of the virus was preserved and reproduced 
intact. How is this possible? 

The DNA-agar technique again provided an answer. With exquisite 
precision, experiments were performed to find out not only whether the 
lambda phage genome was homologous with the K12 Escherichia coli genome 
but also which specific parts of the phage DNA are homologous. These 
precise experiments were made possible by the techniques that A. D. 
Hershey, Director of the Genetics Research Unit, has developed for frac- 
tionating a virus chromosome. From the experiments it was concluded that 
many noncontiguous segments of the DNA in the lambda genome are 
homologous to the DNA of the bacterial host. Furthermore, these segments 
are distributed throughout the entire length of the lambda DNA. This 
complementarity may account for the survival of the viral genetic material 
in its host generation after generation, even though no virus particles appear. 

The results from the lambda virus experiments encouraged an attack on 
another problem more directly connected with tumor appearance. For a 
number of years the polyoma virus has been much studied as one of the 
best known tumor-inducing agents. It has been shown to cause tumor 
formation in a variety of rodent species and even in cell tissue cultures from 
rodents. (See plate 5.) Although the association of the polyoma virus with 
tumor production has been indubitably established, a significant proportion 
of the tumors themselves appear to be virus free. At the same time they still 
contain specific polyoma-induced antigens. Thus the presence of the viral 
genetic material itself within the tumor was strongly suspected. 

DNA-agar experiments were therefore used to test the complementarity 


of the DNAs of the tumors and of the virus. The work was done in coopera- 
tion with the Laboratory of the Biology of Viruses of the National Institutes 
of Health. The findings were that the tumors do indeed contain poly- 
nucleotide sequences that are complementary to those of the polyoma virus 
DNA. This finding is a direct demonstration of the presence of cancer virus 
material in the host it has infected. The implication for understanding the 
biological basis of other kinds of mammalian tumor formation is clear. The 
absence of an inducible virus in a tumor or transformed cell can no longer 
be considered evidence excluding a tumor-causing virus as the agent. The 
application of the DNA-agar technique to a systematic study of the 
complementarity of other tumor DNAs and the DNAs of suspected viruses 
may very well be a long step toward dissipating further some of the mys- 
teries surrounding tumor origin. 

Another long-standing problem was attacked during the year by means 
of the DNA-agar technique. This is the process of cell differentiation; that 
is, how do specialized cells like the heart, nerve, kidney, and liver cells 
develop from the individual fertilized egg with which each organism 
commences? Does each specialized cell of later stages contain the full genetic 
message for the whole organism, or has the specialized cell received only a 
part of the total message contained in the fertilized egg? This has long been 
one of the great problems in developmental biology, familiar in previous 
reports as the field of the Institution's Department of Embryology. 

As has long been suspected, the DNA of a specialized cell was demon- 
strated by the experiments to be identical with the DNA of the germ cell. 
It has the entire genetic message, whether it be heart, liver, kidney, nerve, 
or other type of cell. On the other hand, the so-called "messenger-RNA" 
that takes its pattern from the DNA, and then determines the protein 
content of the specialized cell, was shown not to contain complete matched 
copies of the DNA complement. Parts of the genetic message present in the 
DNA have been suppressed, and the activity of other parts has been 
intensified or repeated, in the specialized cell. The synthesis of specialized 
proteins and consequent enzyme functions therefore ensues for each of the 
specialized cells. The mechanism of cell differentiation within organisms 
therefore is traceable to the differential transcription of the genie material 
in each cell rather than to a distribution of different DNAs among the 
different types of cells. The Biophysics Section's technique therefore proves 
to have another fruitful series of applications in what its staff members call 
"a semiquantitative assessment" of the expression of DNA in different cell 
systems of any animal. Others will agree that the use of the Section's method 
"to detect differences in gene expression during morphogenesis, aging, virus 
infection, or hormonal stimulation presents an exciting challenge." 

This challenge was immediately taken up in part by the Department of 
Embryology in its work during the year. 


Studies of Molecular and Cellular Aspects of Development, the 
Department of Embryology 

Within the Department of Embryology there is a group not unlike the 
experimental petrology group, described above, in its informal organization 
but still coordinated approach. This group's interests center on the molec- 
ular and cellular aspects of cell development. There are a similar agreement 
on the general direction of a search and similar mechanisms for communica- 
tion, and at the same time similar individuality in pursuing specific paths 
of experiment. Besides J. D. Ebert, Director of the Department, the staff 
group this year included D. D. Brown, D. W. Bishop, I. R. Konigsberg, 
and R. L. DeHaan. J. B. Gurdon, at Oxford University, participated during 
the year in a long-range but highly effective collaboration with Brown that 
will be reported upon below. M. E. Kaighn and M. C. Reporter were 
assistant investigators; and the group included the following Fellows: 
H. Denis (Nato Fellow), I. B. Dawid, I. A. Ajdukovic, H. Tiedemann, 
G. C. Rosenquist, 0. Ramirez-Toledano, and F. Beck. 

Ebert describes the primary objective of his Department as the study of 
"interacting systems in development" for biological organisms. In the 
molecular and cellular studies group attention has been centered on the 
inner controls of cells and their interactions with neighboring cells and their 
microenvironment. Like the research topics of the Biophysics Section, these 
interests have partial parallels in a great many university and college biology 
departments or institutes. The situation is much changed from the time 
three or more decades ago when the Department was one of the very few 
centers of embryological research in the world. Nevertheless, its scope and 
penetration are still shared by only a few around the world. 19 

Being interested in systems within the organism, the molecular biology 
group has sought probes at as many different stages and levels as current 
techniques and knowledge made reasonable. A number of them concern the 
genetic material and the initial direction that it gives to the formation of 
the differentiated cell. This is the subject on which the Biophysics Section 
experiments presented some interesting general evidence. But, as Ebert 
wisely notes in his report this year, these events, which have given such 
interesting results within the last year or two, end with ribosomal mediation 
of protein formation. What of the developmental steps that lie beyond the 
ribosomes? He suggests that ten years in the future the important research 
questions will concern the ordering of the specific protein products that 

19 In the United States the following may be mentioned: Johns Hopkins (with which the 
Carnegie Department of Embryology is closely associated), Yale University, Western Reserve 
University, Stanford University, University of California (Berkeley), Massachusetts Institute of 
Technology, Brown University, Princeton University. There are also a number of embryology 
laboratories abroad, including one in nearly every European country. Great Britain seems out- 
standing in this field, but Belgium, Italy, Czechoslovakia, and Japan all have fine work in one or 
more of the fields in which the molecular group of the Carnegie Department is interested. 


"roll off the ribosomal assembly line." How are similar, and even identical, 
protein products fashioned into totally different structures within an 
organism? And how are certain similar, if not identical, cells predestined by 
still unknown properties to totally different functions? It is in anticipation 
of these important problems of the future that the group has looked beyond 
the nuclear-ribosomal relations. Their work of the year displayed a nice 
balance between research on cellular events with direct genetic connections 
and the investigation of cell interaction. 

Because they are a sequel to the DNA homology experiments of the 
Biophysics Section, it may be appropriate to mention first the DNA 
homology experiments conducted by Denis of the group, even though his 
findings must still be considered tentative. Denis's materials were the DNAs 
and RNAs of the frog Xenopus laevis, an animal found to be highly useful 
in other experiments at the Department. First, Denis showed that DNA 
from a Xenopus embryo was homologous with DNA from the liver cells of 
an adult frog; second, he confirmed earlier findings by Brown and Littna 20 
that two of the three classes of RNA are produced by Xenopus cells during 
embryonic development. One of these classes of RNA is thought to be 
messenger-RNA. The messenger-RNA increases rapidly in production 
during early embryonic stages and reaches its maximum when the embryo 
is about two-thirds formed. Finally, he found in " competition" experiments 
that RNA isolated from progressively later stages in embryonic and tadpole 
development is increasingly efficient in competing with adult labeled RNA 
hybridizing with adult DNA. These results would seem consistent with 
those of the Biophysics Section. They further suggest that the PtNAs of 
different stages of development have distinct differences, one from another. 

Another study of the group attacked what Ebert calls an "old and 
especially troublesome question." For a number of years it has been known 
that eggs of several species of animals contain more DNA-like material than 
is found in the complementary sperm. There has been a great deal of 
speculation about the meaning of this apparent anomaly, but the excess 
DNA itself had never been isolated or characterized. It had been loosely 
called "cytoplasmic DNA." During the year Dawid succeeded in isolating 
this elusive DNA from the eggs of Xenopus and those of another frog, Rana 
pipiens. He found the eggs to contain an amount of DNA 100 to 300 times 
greater than that found in the body (somatic) cells of the same species. 
Dawid showed that this excess DNA closely resembles the liver DNA of 
the same species. 

Dawid further attacked the question of what the possible function of the 
excess DNA might be in the developmental process. His first experiments 
suggest that the familiar speculative answer is not correct, namely, that 
the "preformed" DNA is used directly in the formation of chromosomal 

20 Journal of Molecular Biology, 8, 669-687, 1964. 


DNA in early embryonic stages. Perhaps the most fascinating part of the 
question still remains, but the isolation and characterization of this unusual 
DNA constitute a commendable achievement. Even though it seemingly 
concerns a detail, this DNA anomaly is one of the many clear warnings of 
the complexities that may be anticipated in achieving an understanding of 
even the first cell of any organism. 

The relation of fine structure to function within the living cell is a subject 
of continuous interest to both cytology and molecular biology. Even infer- 
ences are difficult, and incontrovertible evidence has been rare. Many 
studies, for example, have implicated the organelles known as nucleoli 
within the nucleus as the site of the synthesis of the ribosomal particles of 
the cell. As was noted previously, the ribosomes are concerned with protein 
synthesis. In a rarely productive long-distance collaboration, Brown, of the 
molecular biology group, and Gurdon, of Oxford University, conducted a 
series of experiments that gave compelling evidence of the synthesis of 
ribosomal RNA within the nucleoli. The materials they used, again from 
Xenopus, illustrate the importance of the scientist's alertness to abnor- 
malities or the unusual in the design of his experiment and in his observation. 

In 1958, Elsdale, Fischberg, and Smith, then at Oxford, described a 
mutant of the Xenopus frog that had only one nucleolus in its cells instead 
of the normal two. The mating of two single-nucleolus frogs produces 
embryos having two, one, or no nucleolus in the cells. Both the double- and 
single-nucleolus animals developed normally. The animals that have no 
nucleolus in their cells develop normally until after the tadpoles are hatched, 
and then soon die. This happens to be the stage at which the large-scale 
formation of new ribosomes begins. Brown and Gurdon showed, through 
their experiments, that the anucleolate frogs failed to form ribosomal RNA, 
although they synthesize DNA, messenger-RNA, and soluble RNA. It is 
interesting that cells having only half the normal complement of nucleoli 
still function in a completely normal manner and allow normal development, 
thus revealing another of the redundancies so characteristic of life systems. 
Another notable one in the nucleus, of course, is the double-strand structure 
of DNA. 

Explorations of the interaction of cells in differentiation are somewhat 
more distant from the broadening base of demonstrated facts about cell 
interiors. The search therefore must be carried on with many fewer reference 
points and on a disconnected trail. Nonetheless, significant progress has 
been achieved, at least in charting the mechanics of cell movement as 
differentiation proceeds in a developing organism. Because they are early 
distinguishable in embryogenesis, heart cells have been a favorite material 
of the group for such study. 

During the year DeHaan added to his earlier observations of the move- 
ments of clusters of precardiac cells. Studying chick embryos, he found that 


these cells follow directed routes of migration into the midline site of begin- 
ning heart formation. His observations in previous years suggested that the 
substratum of cells through which the precardiac cells moved gave some 
directional guidance, or a "path," for them. Conducting experiments during 
the year in which precardiac cells were paired with background (endoderm) 
cells, and other experiments in which the background endoderm was 
disrupted, he showed clearly that the precardiac cells do migrate along a 
definite path to the site of beginning heart formation. 

One of the most interesting aspects of the experiments is that they seem 
to give convincing evidence against a chemical gradient as the means of 
orienting the movement of precardiac cells. DeHaan has suggested a possible 
nonchemical hypothesis to account for movement of the cardiac cell cluster. 
He postulates that the moving exploratory filopodia 21 of the migrating cells 
"sense" the directional information built into the substratum through 
which the cells move. His hypothesis depends on recognition of the known 
differential adhesiveness that characterizes the relation between individual 
cells and their substrata. If confirmed it would provide a fascinating new 
view of cell relations. 

Other experiments of DeHaan's took up the interesting problem of 
"pacemaker" cells of the heart. Their results confirmed an hypothesis on 
which there has been some previous evidence : the heart is composed of two 
different types of cells. DeHaan's data indicated that at best only 25 to 30 
per cent of the cells comprising an embryonic heart can beat spontaneously 
or act as "pacemakers" stimulating the beating of other heart cells. It 
appears that the remaining 70 or 75 per cent, or the vast majority, of heart 
cells cannot generate their own beating stimulus. Thus the pacemaker cells 
appear to be the "heart" of the heart. But as always in science there remains 
an absorbing question beyond the last considered: What generates the 
bioelectric impulse to beat within the pacemaker cell itself? 

The Genetics Research Unit 

The Genetics Research Unit, whose collaboration with the Biophysics 
Section has been mentioned above, has two major research interests. One is 
B. McClintock's study of the control system regulating the somatic expres- 
sion of the genes. The second is the study of the relation between DNA 
structure and chromosomal function by the biochemical genetics group. 
Besides A. D. Hershey, Director of the Genetics Research Unit, the bio- 
chemical group this year included E. Burgi, G. Mosig, and F. R. Frankel, 
as Associates; and E. Goldberg and N. Ledinko (United States Public 
Health Service), as Fellows. 

21 Filamentary protrusions of the cell cytoplasm that can extend, retract, and otherwise move. 
They form part of the cell surface. 


The Biochemical Genetics Group 

The biochemical genetics group belongs to a different type of explorers' 
community within the Institution from those previously described. In this 
group, research activity follows a general pattern clearly inspired by one 
senior member of the staff, even though the conception, execution, and 
interpretation of experiments may be the product of several individuals. 
The photosynthetic research group at the Department of Plant Biology, 
Stanford, California, is similar. Both groups are relatively small, but each 
has had a distinguished history of remaining year after year on the growing 
edge of its special field. 

In the remarkable introduction to his report this year, Hershey describes 
the intellectual environment within which he and those of his group have 
worked for the last decade. It affirms a deep commitment to having as many 
competent explorers as possible, wherever they are to be found. His ideal, 
as he states in the last sentence of his introduction, is the situation where 
laboratories are "encouraged to multiply as fast as the multiplication of 
dedicated research workers permits." 

The same statement might have been made by any one of several other 
research groups in the Institution as well, for that feeling has been part of 
its spirit for more than sixty years. 

The story with which Hershey and his colleagues are concerned, he 
explains, goes back to the now famous twin helical model of DNA developed 
by Watson and Crick at Cambridge University. The basic importance of 
this concept for molecular genetics is now universally recognized. Not so 
widely recognized is the importance of certain discoveries in experimental 
technique. One example was the discovery by Marmur, Doty, and their 
colleagues, at Harvard University, that the complementary strands of the 
DNA helix could be separated and then rejoined by thermal means. Among 
other results this opened the way for the development of the DNA-agar 
technique whose invention and exploitation by the Biophysics Section have 
been recounted previously. A second example was the development, by 
Meselson, Stahl, and Vinograd of the California Institute of Technology, of 
the equilibrium density-gradient centrifuging in solutions of cesium chloride 
and other salts. This technique permitted the first demonstration that 
genetic recombination can occur by means of the fracture and rejoining of 
DNA molecules. 

Hershey then recounts some of the key discoveries about genetic material 
as interpreted from study of the phage chromosome, including his own 
biochemical conclusion that a phage particle contains only one piece of 
DNA. Kaiser and Hogness of Stanford University reached the same con- 
clusion by means of biological assay methods. Marmur of Brandeis Uni- 
versity has shown that only one strand of phage DNA produces the comple- 
mentary RNA that is assumed to direct the synthesis of phage proteins. 


This is the denser pyrimidine-rich strand. 22 Sinsheimer of California Insti- 
tute of Technology discovered a phage containing a single-stranded DNA 
having a ring structure. This was further confirmation of the biological 
competence of a single strand and the first indication that a DNA molecule 
could be circular. Zinder, of the Rockefeller Institute, discovered a bacterio- 
phage from which DNA is completely missing; the genetic material, as in 
many viruses that infect animals and plants, is entirely RNA. Study of the 
effects of radiation on bacteria and phage has revealed that damage to the 
genetic material can be repaired within cells where two-stranded DNA is 
present. This discovery came about as a result of a series of experiments at 
widely different locations, including those by Wacker of Frankfurt, Ger- 
many; Beukers and Berends of Delft, the Netherlands; Hill of Columbia 
University; Setlow and Carrier of Oak Ridge; and Boyce and Howard- 
Flanders of Yale University. In sum, a picture is emerging that shows the 
stability of genetic material to be preserved in spite of continual separation 
and fragmentation of DNA strands during replication. Furthermore, the 
duplication of information in the two strands protects the genetic material 
against accidental damage and error, and operates in still other ways to 
ensure genetic survival. 

Hershey is not disturbed by the striking diversity of genetic mechanisms 
that the phage study of his own and other laboratories has brought out. 
Although it is obvious that the overwhelming preponderance of DNA as a 
genetic material in the biotic world must mean that it is superior for its 
purpose, Hershey observes that "different organisms exploit different ways 
of doing things for a reason best described as historical accident." However, 
there may remain both taxonomic and evolutionary significance in these 
striking differences of the genetic material of one of the simplest of all 
known life forms. 

This year Hershey and his colleagues began to examine some questions 
related to the principle of structural redundance in DNA. They note that 
some of the uses of redundancy call for minor modifications of the classic 
Watson-Crick helical structure. They found one of them during the year 
in their study of lambda phage. 23 Their earlier experiments had shown that 
the ends of lambda DNA molecules can be joined together, by suitable 

22 Pyrimidine is a compound of the general form : 





Cytosine and thymine among the DNA bases are pyrimidines. 

23 Bacteriophages, or phages, are viruses that infect bacteria. Lambda phage, like most of those 
studied in laboratories, infects the colon bacillus Escherichia coli. 


thermal treatment, to form rings or multimolecular threads. The conditions 
required to bring about these changes suggested that complementary 
nucleotide sequences lying at either end of the molecule made the joining 
possible. This hypothesis has now been confirmed by Burgi, chiefly through 
enzymic analysis of structure. Apparently the finished DNA molecule found 
in phage particles consists of a conventional double-helical structure modi- 
fied by short, single-stranded ends with complementary base sequences. If 
the hypothesis is accepted, a model of the interconversion of rings and 
multimolecular threads under thermal treatment may be depicted (fig. 5A). 

A B 

Fig. 5. Two diagrammatic models representing phage structure. A. Probable structure of the 
ring and open forms of phage lambda DNA. Arrows on the ring strands indicate the position of 
enzymic cuts that precede the formation of a finished molecule from a precursor. B. Hypothetical 
paired structure formed by two complementary phage T4 DNA fragments, each fragment of two- 
thirds normal length. 

Finished, biologically competent DNA molecules may be formed from a 
circular precursor by enzymic cuts (fig. 5A). The enzymic cuts and comple- 
mentary nucleotide sequences provide a general mechanism for fragmenting 
and rejoining of DNA molecules, a sort of molecular synapsis 24 that may 
be explained by known forces. 

In collaboration with Cowie, of the Biophysics Section, Hershey was able 
to show further that each of the halves of the lambda DNA molecule 
contains genetic sequences absent from the other. These findings were the 
basis for further experiments on the homology of lambda DNA and its 
bacterial host, as reported in the discussion of the Biophysics Section above. 

The biochemical genetics group also notes from their experiments another 
remarkable feature of lambda phage DNA structure. The guanine-cytosine 
base pairs are noticeably concentrated in the left one-third of the molecule. 
Such a distribution differs strikingly from that of other microbial DNAs, 

24 Synapsis is the joining together of complementary strands of genetic material. 


where the guanine-cytosine pairs are distributed more uniformly. Hershey 
suggests that this may mean diverse historical origins for different parts of 
the lambda DNA molecule. Such differences, he says, could have been 
preserved only if there were a corresponding functional differentiation 
among the parts of the molecule. The finding is consistent not only with 
previous genetic evidence that genes of related function tend to occur in 
clusters within the chromosomes but also with the general chronologic model 
of the evolution of genetic material suggested by the Biophysics Section. 
The model assumes that secular changes must take place by alteration of 
blocks of genetic material rather than by "point" changes. 

Work during the year with T4 phage particles by Mosig provided further 
clues to the mechanism of genetic recombination. Mosig found that a few 
of the T4 particles contain molecules of only two-thirds normal length. 
When isolated these particles prove to be individually noninfective, but two 
or more of them acting together can infect a bacterium. When this happens 
the offspring particles contain DNA molecules of normal length and show 
genetic markers derived from more than one of the parents. According to 
Mosig's interpretation a single particle with two-thirds-length DNA is 
noninfective because parts of its chromosome are missing. In different 
particles different parts are missing, so that two two-thirds-length particles 
can often undergo genetic recombination to produce a complete chromo- 
some. A molecular model for such recombination can also be constructed 
(fig. 5B). 

The findings of Mosig's experiments have some interesting implications. 
First, recombination in this instance clearly involves the joining of two 
DNA fragments near their ends reminiscent of the physically demonstrable 
joining of lambda DNA fragments. Second, several features of recombina- 
tion between the two-thirds-length particles are also seen in genetic recom- 
bination for full-length particles, suggesting that the latter also involves 
rejoining of paired, homologous, molecular ends. Third, the length of DNA 
molecules in T4 phage is a species character under independent gene 
control, not determined directly by the length of the parental molecule. 
This fact was confirmed by experiments using single viable phage particles, 
selected to contain a DNA molecule longer or shorter than the average. 
Within a requisite time their clones of descendants eventually reestablished 
the normal length of molecule. Further insight thus was gained on the 
probable universality of mechanisms controlling the expression of the genes. 
If even the simplest life forms have such exquisite controls the wonderful 
complexities remaining to be discovered in biology are awe-inspiring. 

Experiments during the year by N. Ledinko gave evidence of the presence 
of a rare base, 5-methylcytosine, when lambda phage is grown on a par- 
ticular strain of its host Escherichia coli. This finding adds to the growing 
evidence that rare bases occur in many nucleic acids. The simple picture of 


the classic model (guanine-cytosine, adenine-thymine, or guanine-cytosine, 
adenine-uracil) is not adequate. What functions do these rare bases serve? 
Hershey speculates that they may possibly be associated with the important 
"cutting points" of the single strands. Furthermore, he sees in the host- 
induced modification of the phage u a curious mechanism of speciation, in 
which chromosomal structure is determined by physiological means." The 
possibilities raised by this phenomenon apply to a most significant part of 
the life process. In the best-studied examples, the data suggest that one or 
two phage genes control enzymes that interact with the metabolic system of 
the bacterium and modify entire DNA molecules. Hershey says, "It is now 
clear that analogous phenomena are widespread and diverse, suggesting 
not only a device for establishing genetic barriers but also a possible means 
for controlling gene action in a single species." 

The Photosynthesis Group, the Department of Plant Biology 

The photosynthesis group of the Department of Plant Biology, of which 
C. S. French is the Director, has chosen for its subject one of the most 
important but at the same time one of the most refractory of all the prob- 
lems in biology. Nothing is more basic to the interests of the human race, 
and nothing more challenging to all varieties of curiosity, from the amateur 
nature student to the most esoteric scientist. Indeed, photosynthesis has 
been studied scientifically for more than 350 years, and it is a major subject 
of research in about sixty institutions scattered about the world. Twenty- 
four of them are in the United States and Canada, and twenty-six in western 
Europe and Great Britain. At least eight institutions in the Soviet Union 
are examining photosynthesis from several approaches. 25 Yet, in spite of all 
this activity and the length of time that photosynthesis has been contem- 
plated scientifically, the heart of the system or systems that convert light, 
water, and carbon dioxide into the basic materials of plant nutrition has 
not yet been described adequately. The failure is due in part to the com- 
plexities of the system or systems, but it is also in part due to the stubborn 
refusal of known components to reveal much in vitro. Many answers can be 
sought only in the living cell. 

French, in another excellent report introduction, describes the conceptual 
framework and the intellectual environment within which the research of 
this group is undertaken. The photosynthetic system may be described 
very roughly as beginning with light itself, the wavelengths of which have 
differential values in photosynthetic activity. Next are the pigments, 
including chlorophyll, that absorb the light and supply the driving power 
for the chemical system leading to plant products. This driving power 

25 C. S. French, Photosynthesis, in This Is Life, edited by W. H. Johnson and W. C. Steere, 
pp. 34-38, Holt, Rinehart and Winston, New York, 1962. 


transforms a number of identified and hypothesized substances. The inter- 
actions of these substances lead to oxygen evolution from the plant, and the 
reducing power 26 inside that is associated with the production of high-energy 
compounds for further use in the plant. The last part of this system is now 
reasonably well understood, beginning with the reduction of carbon dioxide 
made possible by the intermediate products from the photochemical action 
of the pigments. However, in comparison, the photochemical part of the 
system still has many mysteries. The chief problems of research now center 
on the light-absorbing system and the products that precede the reduction 
of carbon dioxide within plants. 

The pigment complexes of the photochemical system are now accepted 
as falling into two groups, known as system I and system II. System I is 
composed mainly of a long-wavelength form of chlorophyll a absorbing red 
wavelengths (C a 683). This system seems to act by passing its light energy 
on to a substance known as P700, probably another form of chlorophyll a. 
System II is composed mainly of a group of pigments known as the accessory 
pigments. They absorb light at shorter wavelengths than the pigments of 
system I. The end of the energy-transfer process in system II seems to be 
chlorophyll a 673, also a red-absorbing pigment. Besides the chlorophylls, 
some of the substances thought to be directly connected with or not far 
removed from primary light reactions are cytochromes, plastoquinones, 
ferredoxin, and plastocyanin. There also are detectable substances of 
unknown chemical composition, such as a substance changing light absorp- 
tion at wavelengths near 518 nuz and a substance quenching chlorophyll 
fluorescence. The two systems in operation show an "enhancement" effect; 
that is, the product of the two operating jointly is greater than the sum of 
the outputs of the individual systems operating alone. 

A situation like this, of course, calls for a good model scheme of reactions, 
but, as French observes, even though there have been many heroic attempts 
the half-life of the average good hypothesis in this field seems to lessen with 
each advancing year. Generally hypotheses do not meet the "test of 
predicting the photochemical systems' response to conditions other than 
those used to derive the model and to evaluate its constants." Nevertheless, 
having proper regard for the careful construction of equations and numerical 
data from experiments on the same plant material, it should be possible 
eventually to construct such a model. The present investigations of the 
group are planned in such a way that they may provide the basis for models 
of parts of the complete system at a later date. French summarizes his 
philosophy of operation as he adds, "Thoroughly established relationships 
between a few steps may have more lasting value than even the best 
attempts to form an all-inclusive picture of the process on the basis of our 
present knowledge." 

26 Capacity to add an electron to a compound. 


The other part of French's philosophy of operation is equally important. 
Although the immediate objective must be a limited one, he characterizes 
the present era of research on photosynthesis as one marked by "the efforts 
of many laboratories to fit together into a coherent scheme many phenomena 
observed . . . from different viewpoints." Like research groups in other 
subjects previously described in this review, the photosynthesis group keeps 
in close communication with other laboratories investigating subfields or 
using techniques pertinent to the group's current work. As for so many 
other subjects, this community is world wide. At least five members of the 
biophysical research group at the University of Utrecht, the Netherlands, 
have been in residence at the Department of Plant Biology. At least four 
members of the Department, in turn, have visited Utrecht to exchange 
ideas and observe techniques. Here, as elsewhere in the Institution's 
laboratories, there is a consistent effort to be immediately conversant with 
a worldwide network of facts and interpretations long before they appear 
as published works. 

Besides French, the group during the year included J. S. Brown, D. C. 
Fork, and J. H. C. Smith. Y. de Kouchkovsky, C. J. Soeder, and W. E. 
Vidaver participated as research fellows; and Govindjee, R. Govindjee, 
W. Menke, B. Soeder, and D. A. Webster were in residence for various 
periods as visiting investigators. Most of the lines of investigation of this 
group centered on a measurement of reaction rates in living cells subjected 
to different light exposures and other environmental conditions. A key 
indicator was considered to be fluctuations in rate of oxygen evolution 
during the first few minutes of plant illumination. Thus far much of the 
analysis of oxygen evolution has been largely qualitative. The year's work, 
therefore, was a step toward more precise understanding in quantitative 
terms of a component certain to be used in a later model of the photo- 
chemical subsystems. 

A position for one piece in the prospective model was determined during 
the year in research undertaken on the copper-protein plastocyanin. 
Plastocyanin was discovered some years ago by Katoh, in Japan. It occurs 
only in the green parts of plants, and contains most of the chloroplast 
copper. Chemical experiments have shown it to be capable of reversible 
oxidation and reduction. 27 Until this year it was assumed that plastocyanin 
must be a part of the photosynthetic system, but no direct evidence had 
been obtained for its inclusion. Fork and de Kouchkovsky succeeded in 
establishing a relation between plastocyanin and both pigment system I 
and pigment system II. Previous experiments by others had shown a 
probable oxidation-reduction position between the two systems for cyto- 
chrome/and cytochrome b. It was also thought that an unknown compound 

27 Oxidation is the subtraction of an electron (negative charge) from a compound ; reduction is 
the addition of an electron. 


displaying absorbance changes at 518 niju might have a similar position. 

By experimenting with light exposures in the red and far-red wave- 
lengths, and measuring changes in the light absorption of plastocyanin in 
the cells, Fork and de Kouchkovsky demonstrated not only that plasto- 
cyanin does have a position in the photosynthetic system related to both 
pigment systems but also that pigment system I oxidizes it in the living 
chloroplasts and system II reduces it. This effect was shown for the marine 
alga Ulva lobata and the fresh-water alga Chlorella vulgaris as well as for a 
number of other plants. All the effects were found to be approximately the 
same. They concluded that it was reasonable to assume that plastocyanin 
has a place in the photosynthetic system between the two light reactions 
along with the cytochromes. In addition, they showed that the unknown 
compound absorbing the 518-m/x light is associated with system II but that 
it does not have a position as an intermediary between the two systems. 

Vidaver and French also studied the alga Ulva with particular attention 
to measuring oxygen evolution and uptake in its response to wavelengths 
representing the two pigment systems, intensity of illumination, and dura- 
tion of exposure. The events revealed by their experiments may be described 
about as follows: Algae acted upon by far-red light (activating pigment 
system I) produce a highly reducing substance, which absorbs oxygen both 
during and after far-red-light exposure. French and Vidaver assume that 
the substance responsible for the oxygen uptake is one of the photosynthetic 
intermediates. Concomitantly an oxidized product also results from the 
action of system I. When exposed to red light (activating pigment system 
II) the oxidized product of system I is used to evolve oxygen. The experi- 
ment showed that what may have been assumed to be photooxidation 
(i.e., oxygen uptake directly caused by light exposure) may not necessarily 
be that process. The uptake may be caused by an intermediate chemical 
reaction rather than by the direct action of light on one of the pigment 
systems. An interesting part of the results exhibited in the curves con- 
structed by French and Vidaver for the report are the sensitive and occa- 
sionally unusual responses to light intensity and duration of exposure. Their 
main significance, however, lies in charting some part of the pattern of 
kinetic relations between the pigment systems. 

During the year, Brown, with the assistance of J. Duranton of the Centre 
d'Etudes Nucleaires, Saclay, France, contributed to the differentiation of 
chemical substances in the pigment systems. By means of an anionic 
detergent, sodium dodecyl sulfate, Brown and Duranton partially separated 
the forms of chlorophyll a that have absorption bands at 670 and 680 nux. 
They found that chlorophyll a is a single pigment but that it is attached to 
two different types of protein materials. They did not attempt to analyze 
the difference between the two proteins. Their interim conclusion is that the 
amino acid composition of the two may not be the same, or that it may be 


the same in a different molecular configuration. A third possibility is that 
the chlorophyll a may be bound in different ways to identical proteins. 

The above descriptions of research undertaken with the Institution, both 
by individuals and by the informal research groups, are by no means a 
complete account. Neither are they intended as a selection of the most 
significant results achieved within the Institution's laboratories during 
1963-1964. Instead, they are meant to be representative of the activities, 
thoughts, and manner of scientific life in the six Departments. Indeed, the 
work of some of the Institution's illustrious Staff Members has not hitherto 
been mentioned here. The review might have contained, with equally good 
reason, an account of E. M. Ramsey's continuing and pioneering studies of 
physiological systems of the mammalian uterus at the Department of 
Embryology. It might also have presented additional comment on the 
unique studies of Barbara McClintock on genetic control systems in maize 
at the Genetics Research Unit. Her research on the manner of metabolic 
timing in maize is of the deepest general significance. This year she opened 
a new horizon of versatility in the genetic elements controlling the timing 
of gene action during development of the maize plant. She showed that a 
distinctive pattern of gene expression in a mature tissue may reflect a 
"presetting" of the regulatory mechanism at the gene locus and control 
phenotypic expression in progeny even where the controlling element no 
longer is detectable. Here again the newly unfolding pattern is even more 
wonderfully complex than any preceding model-builder has ever imagined. 

Or the review might have described the fascinating analysis by Abelson 
and Hare, of the Geophysical Laboratory, of the proteins in molluskan 
shells. Hare and Abelson found that amino acid compositions in these shells 
seem to converge on the primitive ancestral types. Furthermore, the 
diverging lines from the primordial types appear to evolve in similar 
patterns. They have thus added one more demonstration to the now rapidly 
growing recognition of the importance of biochemical techniques in evolu- 
tionary studies. And, to cite a last example, the review also might have 
included mention of the experimental taxonomy group in the Department 
of Plant Biology (W. M. Hiesey, H. W. Milner, M. A. Nobs, and D. M. 
Gates), whose work on leaf temperatures this year gives a very significant 
clue as to why different races of the same species of plant can survive and 
reproduce in vastly different environments. Studying races of the monkey 
flower (Mimulus sp.) they found leaf temperatures to be about the same, 
whether observed at high altitude-low temperature stations, or observed 
under low altitude-higher temperature conditions. All these studies are 
described in full on the following pages of the Year Book. 

It has been an exciting year in the Institution. The work of science is so 
absorbing to those engaged in it that sometimes the milestones appear only 
in low relief. Yet, this year, as we look back on it, has markers that will 


remain in the memory of the staff. Among them was the rapidly developing 
knowledge of the remarkable new quasi-stellar objects, discovery of the 
most distant object yet seen, a revised concept of the structure of our 
galaxy, the very stimulating new biochemical evidence of the wonderful 
order of life evolution, a new insight on the biology of tumor causes, pene- 
trating probes into the molecular structure of genetic material, a view of 
the primordial mantle rock of the earth, and several other significant steps 
toward visualizing what lies beneath the very thin known layer of the 
earth's crust. But even more heartening than the pride that staff members 
of the Institution can justly take in their own work is the way in which 
they have continually identified themselves with the progress and problems 
of a world community of scientists, the "society of explorers" named by 
Professor Polanyi. Few reports in this Year Book omit mention in some 
way of a dedication to this larger view. The results described in this report 
are the product not only of devotion, talent, and insight, but also, as the 
members of the Biophysics Section expressed it, "the product of com- 

Losses . . . 

It is with deep sorrow and regret that I must report the death of our 
distinguished Trustee Robert E. Wilson. 

Dr. Wilson died suddenly on September first of this year in Geneva, 
Switzerland, where he was serving as an adviser to the United States 
delegation at the United Nations International Conference on Peaceful 
Uses of Atomic Energy. The seventy-one years of his life had encompassed 
brilliant achievements as research engineer, as business executive, as public 
servant, and as leader in many activities of broad public concern. 

He was born in Beaver Falls, Pennsylvania, March 19, 1892, the son of 
William Hyatt and Madge (Cunningham) Wilson. He attended Wooster 
College (Ohio), from which he graduated with a Ph.B. degree in 1914, and 
the Massachusetts Institute of Technology, where he received a B.S. in 
Chemical Engineering (1916). 

After a year as research assistant with the General Electric Company he 
joined the staff of the Research Laboratory of Applied Chemistry at the 
Massachusetts Institute of Technology as research associate in 1916, 
becoming director two years later and serving concurrently as associate 
professor of chemical engineering (1919-1922). In 1922 he began his long 
business career with the Standard Oil Company of Indiana as assistant 
director of research (1922-1928), advancing through successive levels of 
responsibility to the position of chairman of the board and chief executive 
officer (1945-1958). 

During World War I, Dr. Wilson served first as a consulting engineer of 


the United States Bureau of Mines (1917-1918) and then as a captain and 
later major in charge of the research division of the Chemical Warfare 
Service (1918-1919). From 1947 until February 1, 1964, he was associated 
with the Atomic Energy Commission, serving as a member of various 
committees, including the general advisory committee (1956-1959), and as 
a commissioner (1960-1964). Dr. Wilson consistently opposed the policy of 
a government monopoly on atomic energy and was largely responsible for 
the recent legislation permitting private ownership of atomic materials. 
Upon his retirement from the Commission President Johnson wrote to him : 
"Your outstanding performance and the high esteem with which you are 
regarded as a scientist, a business man, and a public servant must be a 
source of satisfaction to you as your years of public service come to an end." 

Dr. Wilson's many honors included the Chemical Industry Medal (1939) ; 
the Perkin Medal of the Society of Chemical Industries (1943) ; the North- 
western University Centennial Award (1951); the Sir John Cadman 
Memorial Award of the British Institute of Petroleum Technologists (1951) ; 
the Illinois Society of Certified Public Accountants First Annual Public 
Information Award (1956); the Washington Award (1956); and seventeen 
honorary degrees. He was the author of some 120 technical papers and the 
owner or joint owner of about 90 patents for petrochemical inventions. 

Dr. Wilson was a life member of the Corporation of the Massachusetts 
Institute of Technology, a trustee of the College of Wooster, a fellow of the 
Royal Society of England, and a member of the National Academy of 
Sciences and of the American Philosophical Society. Always much interested 
in education for leadership, he repeatedly wrote or spoke on the importance 
of "stressing ethical concepts" as a basic part of education. "Business," he 
said, "wants in its ranks men with high ethical standards, men with a broad 
general background, with or without the capstone of specialization. It is to 
these men who are mentally equipped to step out of their own particular 
field to deal with problems over a wide area that we are looking for the 
leadership and the socially conscious thinking that we so vitally need 
today." In his view, "We badly need men today with the broad outlook, 
men with a sure grasp of their intellectual heritage, mentally and morally 
disciplined and trained to evaluate data and arrive at sound decisions in 
all aspects of life." 

In 1953 Dr. Wilson was elected to the Institution's Board of Trustees; 
he served as a member of the Executive Committee from 1961 and as a 
member of the Committee on Terrestrial Sciences from 1954 until his death. 
Of science he said, "There is one great thing about the frontiers of science 
that is in striking contrast to our shrinking geographical frontiers — the 
frontiers of science are ever expanding into the unknown. All we need are 
capable men to blaze the new trails and adequate incentives to keep them 
on the job." The loss of such a man will be deeply felt throughout the 
Carnegie Institution. 


Within a month after the close of the report year we also lost three 
outstanding pioneers in the work of the Carnegie Institution. 

Dr. Eugene Thomas Allen, who celebrated his hundredth birthday on 
April 8, 1964, died at Arlington, Massachusetts, on July 19. A graduate of 
Amherst College (1887), he received his doctorate in chemistry from Johns 
Hopkins University in 1892. Thereafter he served in several teaching 
positions until 1901, when he joined the staff of the United States Geological 
Survey. There, under the direction of Arthur L. Day, he began a systematic 
investigation of the isomorphism and thermal properties of the feldspars, 
which he continued at the Geophysical Laboratory after 1907. The early 
results of this work, representing the first attempt to use laboratory methods 
to examine in detail the properties of natural minerals, were published by 
the Institution in 1905. One of Allen's finest contributions as a chemist was 
the group of analyses of a number of metals to determine their purity for 
use as fixed points in high-temperature thermometry. 

During World War I he contributed significantly to methods of analy- 
sis for the manufacture of optical glass and also studied the role of 
arsenic in optical glass. After the war, until his retirement from the 
Laboratory in 1932, most of his work dealt with the emanation aspects of 
volcanic activity. This work included a study, with E. G. Zies, of the 
fumaroles of the Katmai region, the Valley of Ten Thousand Smokes, and 
work with Dr. Arthur L. Day on the hot springs of Lassen National Park, 
the steam wells and other thermal activity in California, and the hot springs 
of Yellowstone National Park. He remained as a Research Associate of the 
Geophysical Laboratory after his retirement until 1936. 

Dr. Allen was elected a member of the National Academy of Sciences 
in 1930. 

Dr. Warren H. Lewis, long associated with the Department of Embry- 
ology, died at the age of 94 in Philadelphia on July 4. A pioneer in the field 
of tissue culture, he was one of the first to use motion pictures in studying 

Dr. Lewis was born in Sufneld, Connecticut, on June 17, 1870. He 
received a B.S. degree from the University of Michigan in 1894 and an 
M.D. from Johns Hopkins University in 1900, where he served at pro- 
gressive levels on the faculty until 1940. At Hopkins he was closely associ- 
ated with the work of our Department of Embryology, becoming a full-time 
staff member in 1919. He had by that time, with Mrs. Lewis, developed a 
method of growing embryonic tissues in artificial media that made it 
possible to observe individual living cells under the microscope. His research 
for the Institution brought about many advances in knowledge and tech- 
niques, which won him wide recognition in his field. 

After his retirement from the Institution in 1939 Dr. Lewis continued to 
work for a year under a grant from the International Cancer Research 
Foundation. He became an active member of the staff of the Wistar Insti- 


tute in Philadelphia in 1940 and Professor Emeritus in 1958. He was the 
editor of the classic Gray's Anatomy from its twentieth to its twenty-fourth 
editions and the author of some 150 papers on anatomy, embryology, and 
cytology. He was a past president of the International Society for Experi- 
mental Cytology, the American Association of Anatomists, and the Mount 
Desert Island Biological Laboratory, and a fellow of the American Associ- 
ation for the Advancement of Science. He was also a member of the 
American Philosophical Society and the National Academy of Sciences. 

Dr. Frederick Hanley Seares, a retired astronomer of Mount Wilson 
Observatory, died in Honolulu, Hawaii, at the age of 91, on July 21. 

Born at Cassopolis, Michigan, on May 17, 1873, he was educated at the 
University of California, where he received his B.S. degree in 1895, served 
as a fellow from 1895 to 1896 and from 1898 to 1899, and was awarded an 
honorary LL.D. in 1930. He also studied at the University of Berlin, 
1899-1900, and the University of Paris, 1900-1901. The University of 
Missouri conferred an honorary LL.D. on him in 1934. From 1901 to 1909 
he was director of Laws Observatory at the University of Missouri, from 
which position he went to be superintendent of the computing division and 
editor of publications at Mount Wilson Observatory in 1909. In 1925 he 
was appointed assistant director of the Observatory, a post he held until he 
retired in 1940 to become a Research Associate. 

Dr. Seares' most important astronomical work dealt with the precise 
measurement of the brightness and color of stars and their distribution in 
the galaxy. In these fields he was an outstanding authority. He also con- 
tributed to the investigation of the sun's magnetic field. 

He was a member of the American Academy of Sciences and the American 
Philosophical Society, an associate of the Royal Astronomical Society, and 
the Bruce Medalist (1930) of the Astronomical Society of the Pacific. From 
1919 to 1938 Dr. Seares was president of the commission on stellar photom- 
etry of the International Astronomical Union. 

Although the Institution lost an outstanding director of the Mount 
Wilson and Palomar Observatories when Dr. Ira S. Bowen retired on 
June 30, 1964, it has gained tremendously in his return to full-time research 
as the first Carnegie Distinguished Service Staff Member at the Observa- 
tories. In this capacity he will work on the design and improvement of 
large telescopes and other problems. 

Dr. Bowen is a graduate of Oberlin College (A.B., 1919). At the Univer- 
sity of Chicago, which he attended from 1919 to 1921, he came to know 
Robert A. Millikan, who invited him to join the faculty of the California 
Institute of Technology as instructor in physics. After receiving his Ph.D. 
degree at the Institute in 1926, Dr. Bowen continued to teach and to do 


research there, advancing to the rank of professor, which he held from 
1931 to 1945. 

His work during the 1920's lay chiefly in the area of spectroscopy. As a 
direct result of his investigation of the spectra of certain ions he was able 
in 1927 to solve one of the most baffling astronomical problems of the day : 
to identify, and explain consistently with the quantum theory, distinctive 
emission lines of spectra from gaseous nebulae. He identified the lines with 
transitions between low-energy levels — possible only under conditions of 
extremely low gas pressures. This achievement, one of the landmarks of 
astronomical research of the decade, provided a fruitful source of informa- 
tion on the physical properties of the gaseous nebulae, including the relative 
abundance of the elements present. 

Beginning in the early 1930's Dr. Bowen was associated with the planning 
of the Palomar project and was responsible for the final testing and finishing 
of the 200-inch Hale telescope mirror. He devised ingenious optical tests to 
determine the corrections to be made in the mirror after its mounting on 
the telescope and personally supervised that difficult operation. He was also 
consulted in the design of several other telescopes, including the 120-inch 
instrument at the Lick Observatory of the University of California and the 
84-inch optical telescope and the solar telescope at Kitt Peak National 
Observatory in Arizona. 

During World War II Dr. Bowen made notable contributions toward 
improving the trajectories of underwater missiles and developing more 
effective military photography. He supervised the photographic section of 
the California Institute of Technology's rocket project for the Office of 
Scientific Research and Development. 

In 1946 Dr. Bowen was appointed Director of the Institution's Mount 
Wilson Observatory, and two years later he became the head of the com- 
bined Mount Wilson and Palomar Observatories, jointly operated by the 
Institution and the California Institute of Technology. 

He has contributed much to the improvement of spectrographs and 
cameras used in astronomical work, and has devised instruments, such as 
the image sheer, to increase the efficiency of spectrographic observations. 
He has also determined in greater detail the spectra of many chemical 
elements, devising vacuum spectrographs that made this work possible. His 
many scientific publications embrace the fields of spectroscopy, the compo- 
sition of gaseous nebulae, cosmic rays, optics, and the design and construc- 
tion of large telescopes. 

Many honors have come to Dr. Bowen for his scientific achievements. 
He was elected to the National Academy of Sciences in 1935 and was 
awarded its Henry Draper Medal in 1943. He received the Potts Medal of 
the Franklin Institute in 1946, the Rumford Medal of the American 
Academy of Arts and Sciences in 1949, the Ives Medal of the Optical 


Society of America in 1952, and the Bruce Medal of the Astronomical 
Society of the Pacific in 1957. He holds honorary degrees from Oberlin 
College, Princeton University, and the University of Lund in Sweden. 

Mr. William F. Steiner, who has been with the Department of Terrestrial 
Magnetism for more than 49 years, retired on June 30. Mr. Steiner came 
to the Department when he was sixteen years old on March 16, 1915, as a 
mechanician apprentice. Through the years he developed such skill as an 
instrument maker that he became chief of the shop section, whose responsi- 
bilities he fulfilled with singular effectiveness until his retirement. During 
the 1930's the great precision of the magnetic instruments Mr. Steiner built 
was vital to the success of the Department's magnetic investigations. In the 
next decade he contributed significantly to the construction of the Institu- 
tion's cyclotron. He also did all the precision work on the prototype models 
of Dr. Bush's automatic microtome and micromanipulator. Mr. Steiner 
stayed with the Institution a year beyond the normal retirement age in 
order to go to Argentina as one of the representatives of the Department 
in the construction of a 100-foot parabola for radio astronomical studies. 

. . . and Gains 

On September 11, 1963, Dr. Philip H. Abelson, Director of the Geo- 
physical Laboratory, received the Gold Medal Award for Scientific Achieve- 
ment from the Chemistry Alumni Association of the City College of New 
York. On June 15, 1964, Yale University conferred upon him the honorary 
degree of doctor of science. On November 19, 1963, J. Frank Schairer, staff 
member of the Laboratory, was presented the Roebling Medal, the highest 
award of the Mineralogical Society of America, for outstanding achievement 
in mineralogy. At the same time Dr. Nobuo Morimoto, a postdoctoral fellow 
at the Laboratory from 1957 to 1959 and again in 1962, now at the Uni- 
versity of Osaka, Japan, received the Mineralogical Society of America 
Award for 1963. He is the first Japanese to receive this award. Professor 
C. E. Tilley, of Cambridge University, a Carnegie Research Associate 
working at the Laboratory, received an honorary doctor of science degree 
from the University of Sydney, Australia, in January 1964. 

At the Department of Terrestrial Magnetism, Dr. Louis Brown, staff 
associate, was one of a team of three scientists awarded the Amerbach 
Prize of the University of Basel, Switzerland. The prize, founded in 1962 
in memory of the Amerbach family, who in the fifteenth and sixteenth 
centuries played an important role at the University of Basel as presidents 
and professors, was given in recognition of Dr. Brown's contribution to the 
design and construction of the first artificial source of polarized deuterons 


and its successful application in nuclear physics research. Dr. Brian J. 
McCarthy, staff member, received an award for scientific accomplishment 
from the Washington Academy of Sciences on January 16, 1964, "for his 
role in deciphering the biosynthetic relationships among nucleic acids." 

Dr. Maarten Schmidt, staff member of the Mount Wilson and Palomar 
Observatories, was awarded the Helen B. Warner prize of the American 
Astronomical Society. This prize is awarded annually, to a person under 
35 years of age, for a significant contribution to astronomy during the five 
preceding years. Dr. Schmidt's paper describing his work on quasi-stellar 
radio sources will be given on December 29, 1964, at the American Associ- 
ation for the Advancement of Science meeting in Montreal, Canada. 

Dr. Louis B. Flexner, Research Associate of the Institution and former 
staff member of the Department of Embryology, was elected to membership 
in the National Academy of Sciences on April 28, 1964. 

Plate 1 

Report of the President 


03 ^ 

5 ^ 

'o ~ 

6 -+' 
O 'O 



Report of the President 


/ ■ . 

88 "^ 

.9 ** 

— 0) 

bJD O 

3 f • 
i — i "_3 on 

oo J2 o 

X o o 


O d o 

H C3 

.2 +2 


Report of the President 

An Sc ("late spiral") type of galaxy, XGC 5457. Our Milky Way galaxy is now interpreted as 
being of the Sc type. 

Plate 4 

Report of the President 

Photomicrograph (30 X ) of an alkali basalt composed of 
olivine and avigite (both gray), plagioclase feldspar 
(white), and magnetite (black). Sample from a prehistoric 
lava flow north of Keauhou, Hualalai, Hawaii. 


Report of the President 

Polyoma -virus-induced tumors in a three-month-old mouse. Bilateral salivary gland 
tumors, mammary tumors, and multiple-hair-follicle tumors are visible. 

National Institutes of Health-National Cancer Institute photograph. 

Reports of Departments 
and Special Studies 

Mount Wilson and Palomar Observatories 

Geophysical Laboratory 

Department of Terrestrial Magnetism 

Committee on Image Tubes for Telescopes 

Department of Plant Biology 

Department of Embryology 

Genetics Research Unit 

Cytogenetics Laboratory 

Mount Wilson and Palomar 


Operated by Carnegie Institution of Washington 
and California Institute of Technology 

Pasadena, California 

Ira S. Bo wen 

Horace W. Babcock 

Associate Director 


Ira S. Bo wen 

Carl D. Anderson 

Horace W. Babcock 

Jesse L. Greenstein 

Robert B. Leighton 

Allan R. Sandage 


Introduction 5 

Observing Conditions 6 

Solar Observations 6 

Large-scale magnetic fields .... 6 

Development of sunspot groups . . 8 

Velocity fields 8 

Eclipse observations 8 

Planets and Comets 9 

Mars 9 

Jupiter and Saturn 9 

Physics of comets 10 

Stellar Spectroscopy and Photometry . .11 

Various spectroscopic investigations . 11 

White dwarfs 12 

Faint blue stars 13 

Absolute spectrophotometric standards 14 

Color systems 14 

Chemical composition of stellar 

atmospheres 15 

Colors and magnitudes 16 

Emission lines of ionized calcium . .17 
Absolute magnitudes from the width 

of Ha 18 

Influence of rotation on colors and 

magnitudes of main-sequence F- 

type stars 19 

U Geminorum stars 19 

Old novae 19 

TTauri stars 20 

Wolf-Rayet stars 20 

Galactic clusters 21 

Globular-cluster photometry ... 22 

Properties of the galactic center . . 23 

Interstellar Gas and Gaseous Nebulae . 23 

Combination of optical and radio data 23 

CN lines 24 

He I nebular absorption 25 

Planetary nebulae 25 

Interstellar matter 25 

Radio Sources 26 

Optical observations of quasi-stellar 

sources 26 

Identification with galaxies .... 27 

Interpretation 27 

Galaxies 29 

Magnitudes and redshifts .... 29 

Investigations of individual galaxies . 29 
Peculiar and interacting galaxies . .31 

Faint features in galaxies .... 32 

Interference-filter photography ... 32 

Polarization 32 

Compact galaxies 33 

Catalogue of galaxies and clusters of 

galaxies 34 

Clusters of galaxies 34 

Supernovae 34 

Southern Sky Survey 35 

Theoretical Studies 35 

Instrumentation 36 

Data acquisition 36 

150-foot solar tower 37 

Prime-focus scanner 37 

Image tubes 38 

Optical systems 39 

Galaxy image synthesizer .... 40 

Diffraction gratings 40 

Southern Hemisphere Site Survey . .41 

Seeing monitors 41 

Chile 41 

Other southern countries .... 42 

Guest Investigators 42 

Staff and Organization 48 

Bibliography 50 

Carnegie Institution of Washington Year Book 63, 1963-1964 


Dr. Ira S. Bowen retired from his ad- 
ministrative duties on June 30, 1964, the 
end of this report year. He had served as 
Director of the Mount Wilson Observa- 
tory since January 1, 1946, and as 
Director of the combined Mount Wilson 
and Palomar Observatories since April 1, 
1948. Under his guidance, the high 
standards established by the first Direc- 
tor, George E. Hale, and continued by 
Walter S. Adams during his administra- 
tion from 1923 to 1946, were rigorously 

The past eighteen years have seen the 
retirement from the Observatory group of 
practically one whole generation of staff 
members, and the building, under Dr. 
Bowen's leadership, of a new and younger 
organization. His administration has 
been marked by the completion of the 
200-inch Hale telescope, to the design 
and perfection of which he personally 
made many fundamental contributions; 
and by the development of policies and 
programs for successfully combining the 
operations of the two large observatories. 

It is a matter of great satisfaction to 
his associates that Dr. Bowen's connec- 
tion with the Observatories is being con- 
tinued with his appointment as Dis- 
tinguished Service Staff Member. 

In Year Book 1 of the Carnegie In- 
stitution of Washington (1902), a com- 
mittee of astronomers, reporting to the 
Board of Trustees, outlined the require- 
ments for choosing the site of a large 
astronomical observatory in the southern 
hemisphere. The first requirement was 
that the observatory should be south of 
the 30th parallel of latitude. It was also 
regarded as desirable that the site should 
offer clear skies, a dry and equable 
climate, and a fair degree of elevation 
above sea level. 

The following year, in Year Book 2, a 
committee consisting of Lewis Boss, 
W. W. Campbell, and George E. Hale 
devoted 165 pages to a proposal for 

southern and solar observatories. This 
proposal included the famous report by 
W. J. Hussey, giving his appraisal of such 
prospective sites as Mount Wilson and 
Palomar Mountain for a solar observa- 
tory. Considerable space was also given 
to a discussion of possible locations for a 
southern observatory in Australia, South 
America, and South Africa. Replies were 
presented from thirteen eminent astrono- 
mers whose views on the proposal to 
establish a new southern observatory 
had been solicited; they were all favor- 
able. The committee discussed at length 
the arguments for such an observatory 
and recommended "that the Carnegie 
Institution should enter this field 
[measurement of the velocities of stars 
and astrophysics] and provide for use in 
the Southern Hemisphere the most 
powerful reflecting telescope that would 
be sanctioned by experience and the 
dictates of common prudence." 

Although the emphasis today would be 
less specifically on velocities and more on 
a wide range of astrophysical problems, 
the recommendation of the Boss Commit- 
tee is, if anything, more valid now than 
when it was made in 1903. The wide- 
spread and growing interest among 
astronomers in the possibilities of large 
telescopic equipment in the southern 
hemisphere is one of the reasons why the 
Observatories, with generous sponsorship 
by the Institution, have undertaken 
within the past year to investigate astro- 
nomical observing conditions at selected 
locations in southern latitudes. All factors 
bearing on the quality of observatory 
sites are being given due attention, but 
emphasis is placed on the quantitative 
measurement of the "seeing" — the deg- 
radation of image quality by inhomo- 
geneity and turbulence of the earth's 
atmosphere. The quality of a site de- 
pends more heavily on this than on any 
other factor, once an adequate number of 
clear nights is assured. Indeed, good 


seeing is fully as important as large 
telescopic aperture. One of the main 
objectives of the site testing now under 
way is to compare the seeing at Mount 
Wilson and Palomar Mountain with that 
at the best prospective locations in 
Chile, Australia, New Zealand, and other 
countries in the southern hemisphere. 

Until quite recently the portable field 
instruments generally employed for site 
testing were small refractors with which 
the observers made visual estimates of 
the seeing from time to time. But the 
conversion of such subjective estimates 
to a reliable measure of the image quality 
that would be found in a large telescope 
under the same conditions is a matter of 
difficulty and uncertainty. 

As was reported a year ago, an ad- 
vanced type of photoelectric seeing 
monitor has been developed. This instru- 
ment utilizes an 8-inch equatorial re- 
flecting telescope with a very stable 

mounting. It is equipped with an optical 
beam splitter and with a crystal-driven 
knife-edge in the focal plane. The instru- 
ment tracks a bright star automatically 
and produces a continuous record of the 
amplitude of image tremor induced by 
the atmosphere, self-calibrated in seconds 
of arc. This tremor amplitude corre- 
sponds approximately to the diameter of 
the "seeing image" that would be de- 
livered by a large telescope. 

Four of these astronomical seeing 
monitors (ASM's) have been built and 
tested locally. They now form the instru- 
mental mainstay of a program of site 
investigation that the Institution is con- 
ducting in the southern hemisphere. Two 
of the ASM's are in Chile; one is being 
operated at Palomar Mountain; and 
another, already used briefly in New 
Zealand and in Australia, is scheduled 
for further operation in Australia. 


Rainfall at Mount Wilson was 23.96 
inches for the year, as compared with a 
60-year average of 34.67 inches. Total 
snowfall was 40 inches. 

Observations made with the major 
telescopes were as follows: 


Number of Number of Total 

Complete Partial Hours 

Nights Nights Worked 

235 56 2344 

264 39 2748 





Routine solar observations were made 
by Cragg, Utter, and Howard on 326 
days. The numbers of records of the 
various kinds made between July 1, 1963, 
and June 30, 1964, were as follows: 

Direct photographs 315 

Ha spectroheliograms, 30-foot focus 1132 

K2 spectroheliograms, 30-foot focus 1153 

Magnetograms 201 

Magnetic classifications of sunspot 
groups were made on 153 days during the 

Large- Scale Magnetic Fields 

The reduction of daily magnetograms 
covering 4}^ years continues. More than 
700 isogauss maps have been drawn, and 
nearly 60 synoptic (rotation) charts 
covering the period have been plotted. 
This work is being done in cooperation 
with the Lockheed Solar Observatory by 
means of a contract with the Advanced 
Research Projects Agency. Dr. V. Bumba, 
Howard, and Sara F. Smith of the 


Lockheed Solar Observatory have con- 
tinued the investigation of the distribu- 
tion of large-scale magnetic fields. It is 
apparent that the large-scale pattern of 
the solar magnetic fields is for the most 
part the result of the spreading out and 
stretching by differential rotation of 
portions of the magnetic fields of old 
active regions. These large-scale features 
persist for many months; the small-scale 
patterns connected with active regions 
can change in a matter of days. The main 
direction of motion of the migrating 
fields is eastward and poleward. The 
following polarity in each hemisphere 
usually predominates in the poleward 
drift of fields. The polar magnetic field 
measurements record this quantized 
migration of fields. These observations 
confirm the earlier finding that, if there 
is an invariant component of a general 
solar field, it can be no stronger than a 
few tenths of a gauss. 

Judging from the distribution of the 
large-scale magnetic fields, an observer 
at a distance, in the equatorial plane, 
would, during some parts of the 22-year 
cycle, observe the sun as a magnetic 
variable star (if it were possible to discern 
variations of a few gauss) with irregular 
fluctuations and reversals in intervals of 
the order of a few days. An observer 
looking along the polar axis would see the 
sun as a reversing magnetic variable with 
a cycle of about 22 years. 

It was found from the study of the 
development of active regions that the 
supergranulation structure plays a very 
important role in the evolution of active 
regions. Similarly, there are indications 
that very large-scale cellular patterns 
exist with diameters of the order of 
400,000 km, outlined by weak magnetic 
fields and calcium emission. Spot groups 
appear to form at the crossings of such 
giant cells. Magnetic configurations con- 
nected with the structure seem to be 
quite complicated. Observations are 
planned to find other physical characteris- 
tics that may clarify these patterns. 

Dr. John M. Wilcox of the University 
of California Space Sciences Laboratory 
at Berkeley, and Howard, have investi- 
gated the solar origin of the interplane- 
tary magnetic field. The photospheric 
magnetic field (measured with the solar 
magnetograph) was compared with the 
field measured in the undisturbed inter- 
planetary medium by magnetometers on 
the satellite Imp (Interplanetary Moni- 
toring Platform). Preliminary analysis 
indicates that a good correlation exists 
between the direction of the field in a 
small area near the center of the visible 
solar disk and the direction of the inter- 
planetary field near the earth, with a sun- 
earth transit time of about 43^ days. 
This suggests that during the present 
quiet part of the solar cycle and in the 
absence of geomagnetic storms, magnetic 
field lines coming from the photosphere 
near the center of the visible disk are 
stretched out in a regular order by the 
highly ionized streaming solar plasma, so 
that they flow by the earth. This observa- 
tional method may provide useful infor- 
mation on the radial configuration of the 
solar magnetic field. 

Studies of solar magnetic fields and 
velocity fields at the 60-foot tower 
continued. Mr. Neil Sheeley discovered 
an interesting and significant inverse 
relationship between numbers of polar 
faculae and sunspot number, and has 
pointed out that the polar faculae, like 
other faculae, are directly indicative of 
the presence of magnetic fields. This 
relationship provides a powerful means of 
extending our knowledge of the behavior 
of the polar fields into the past as far as 
photographic records exist. 

R. B. Leigh ton has developed a simple 
quantitative model, based on a random 
walk, which seems to provide a physical 
mechanism for the dispersal of sunspot 
groups and for the growth and develop- 
ment of magnetic regions on the sun. 
The model accounts quantitatively for the 
rates of growth and the shapes of mag- 
netic regions; the sign, the strength, and 



the time variation of the polar fields; and 
the rate of poleward migration of polar 

Mr. Alan Title has built and tested a 
dual automatic cine camera for use with 
the beam-splitter apparatus at the 13- 
foot spectroheliograph. With it he plans 
to study the time development of 
magnetic and velocity fields over inter- 
vals of many minutes or a few hours. 

Development of Sunspot Groups 

Dr. V. Bumba and Howard studied the 
birth and early growth of individual 
active regions. It seems that the charac- 
teristics of development during the early 
growth of a region are closely connected 
with the preexisting supergranulation 
structure. The first brightenings of the 
new-born calcium plage always occur at 
the crossings of adjacent supergranules. 
The development of the emission then 
follows the intersupergranular space. 
Later development of the large spot 
groups results in the filling-in of whole 
supergranules with emission. The follow- 
ing and leading portions of the emission, 
which are connected with the following 
and leading magnetic polarities, do not 
develop at the same time or at the same 
rate. In general, the following portion 
of the region develops before the leading 
portion. For example, of 43 groups which 
became type C groups on the disk, and 
which definitely started their develop- 
ment on the visible disk between August 
1959 and August 1961, 38 showed this 
type of development; only 4 showed 
development of the leading portion first; 
and 1 was irregular. In all cases for which 
magnetograph data were available, the 
new regions seemed to form in an old 
expanding region of one polarity. Out of 
the regions mentioned above, there were 
good magnetograph data for 24 groups; 
16 of them were in old leading-polarity 
regions, and 8 were in old following- 
polarity regions. During the develop- 
ment of a group, the small sunspots are 

found to occur at the intersupergranular 
space near the crossings of adjacent cells. 

Velocity Fields 

The study of velocity or "Doppler" 
fields on the sun, using the solar magneto- 
graph, continues under the direction of 
Howard. Cragg, Mr. J. W. Harvey of the 
Lockheed Solar Observatory, and Utter 
have made most of the observations. A 
small aperture is held fixed in a region on 
the sun while the Doppler shift of the 
line is recorded on a strip-chart recorder 
and in a digitized form on punched tape. 
The ubiquitous 5-minute oscillation is 
seen — changing from time to time in 
amplitude and phase, occasionally dis- 
appearing for some minutes. The reduc- 
tion of the material is not yet completed. 
From examination of some of the power 
spectra derived from the observations 
it is apparent that in many spectra there 
are, in addition to the obvious peak at 
300 seconds, secondary maxima near 250 
seconds and sometimes at 360 and 400 
seconds. At first it was thought that a 
longer period of about 45 minutes ob- 
served in the autocorrelation plots might 
be physically meaningful, but it seems 
to be a spurious result due to the fact 
that the observations did not cover a 
long enough time. 

Eclipse Observations 

In collaboration with Dr. G. Righini of 
the Astrophysical Observatory of Arcetri, 
Deutsch obtained a spectrogram of the 
corona at the eclipse of July 20, 1963. 
For this experiment they adapted a two- 
prism instrument formerly used for 
stellar spectroscopy at the Cassegrain 
focus of the 100-inch telescope. The 
modified spectrograph was coupled with 
a suitable 6-inch telescope and heliostat. 
The eclipse was observed with this 
apparatus from a DC-8 airplane at 40,000 
feet above Fort Providence, Northwest 
Territories, Canada. The same aircraft 
carried a number of other eclipse ob- 
servers and their equipment. The expedi- 



tion was sponsored jointly by the Douglas 
Aircraft Company and the National 
Geographic Society. 

The slit of the spectrograph was radial, 
with one end at the equator on the east 
limb and the other at a height of 2.2 
solar radii above the limb. To facilitate 
the separation of the F and K components 
of the corona, a Polaroid filter was em- 
ployed to exclude light linearly polarized 
with the electric vector tangential to the 
limb. A plate of good spectral resolution 
was obtained. Besides the familiar fea- 
tures associated with the dust corona (F) 
and the electron corona (K), the spectro- 
gram shows an unexpected strong, sharp 
emission line at the wavelength of the 
K line of Ca II. This feature extends 

from the limb to a height of about 1 solar 
radius. As the H line of Ca II cannot be 
seen in emission, the emission line at K 
must be attributed to an unknown spec- 
trum of some other ion or to an unknown 
process in coronal Ca II that emits H and 
K with a highly anomalous intensity 
ratio. The observed line has not been 
explained, but Deutsch and Righini have 
assembled compelling reasons for be- 
lieving that it cannot be spurious. It 
must be a real and highly anomalous 
feature of the equatorial corona near the 
phase of sunspot minimum. Plans are 
advancing to repeat the experiment, with 
instrumental improvement, from a DC-8 
aircraft based in Tahiti, at the long and 
favorable total solar eclipse in May 1965. 



On the basis of intensity measures of 
lines in the 5*> 3 -C0 2 band at X8600 A, 
discovered last year in the spectrum of 
Mars, and the known strength of the 
strong band at X2.05 /x, a value for the 
surface pressure of Mars of p = 25 dz 15 mb 
has been determined through a curve-of- 
growth procedure carried out by Munch 
in collaboration with Drs. L. Kaplan 
and H. Spinrad of the Jet Propulsion 
Laboratory. This Martian value of p, 
about one-third of previous estimates, 
implies a number abundance for C0 2 of 
nearly 20 per cent. In order to reduce the 
existing uncertainties in the value of p, 
for the forthcoming opposition it is 
planned to remeasure the weak 5*/ 3 
band of C0 2 photoelectrical^ with the 
100-inch coude scanner. The strength of 
individual rotational lines in the strong 
band at X2.05 /jl will be measured also 
by means of a multiplexed chopping 
system and PbS detectors now being 
designed for the 100-inch coude by 
Munch and Dr. G. Neugebauer. 

Jupiter and Saturn 

Munch has continued the photometric 
study of Jupiter and Saturn in collabora- 
tion with Mr. R. Younkin of the Jet 
Propulsion Laboratory. By means of a 
moving entrance diaphragm adapted to 
the Ebert photoelectric spectrometer, 
direct limb darkening curves have been 
obtained at the 60-inch Mount Wilson 
telescope. The use of fairly narrow 
spectral band passes has provided a 
separation of the phenomena taking 
place through purely continuous scatter- 
ing from those introduced by molecular 
absorption. It has been established in this 
way that the appearance of Saturn in 
the photographic infrared is mostly 
determined by the effects of the CH 4 
absorption. Avoiding regions with strong 
CH 4 bands, the contrast between the 
equatorial belt and the temperate zone 
decreases with increasing wavelength. 
This observation suggests that the equa- 
torial belt of Saturn arises from a concen- 
tration of selectively absorbing solid 
particles, relative to the gas producing 



Rayleigh scattering, larger than that 
present at higher latitudes. A similar 
interpretation of the zonal structure ob- 
served in Jupiter is tentatively con- 
sidered, pending the results of calcula- 
tions now being carried out for the 
properties of diffusely reflected light in 
model atmospheres containing diverse 
amounts of aerosols. 

The structure of the atmospheres of 
Saturn and Jupiter is being further 
studied through measures of the in- 
tensities of the CH 4 and NH 3 absorptions 
at various points on the planetary disks. 
The behavior of entire bands has been 
measured photoelectrically with the Casse- 
grain scanner; that of individual lines is 
obtained from spectrograms taken at the 
100-inch coude. The equivalent width of 
the NH 3 lines and some CH 4 bands 
decreases toward the limb of Jupiter, at 
fixed latitude, in a manner revealing the 
pressure dependence of the line absorp- 
tion coefficient. An empirical procedure 
for probing the depth dependence of the 
pressure thus appears possible when some 
laboratory data on strength and collision 
width parameters become available. The 
residual intensity within the X8890 CH 4 
band, the strongest one in the photo- 
multiplier infrared, in Jupiter shows such 
a pronounced decrease from the center of 
the disk toward the poles that tempera- 
ture effects may be playing a role at the 
higher levels of the atmosphere. The limb 
darkening curves of Jupiter at this 
wavelength show sharp reversals right at 
the poles, indicating the presence of very 
high level polar clouds or possibly frozen 
CH 4 . A detailed comparison of the spectra 
of Jupiter and Saturn in the region of the 
NH 3 band at X7900 A and a laboratory 
spectrum of CH 4 in the same region ob- 
tained by Phillips at Berkeley has shown 
that the identification of ammonia in 
Saturn reported by Dunham is probably 
unfounded unless its presence in gaseous 
form is time dependent. 

The calculations of the absorption 
coefficient arising from the pure rotation 

and translation pressure-induced dipole 
transitions of H 2 , for wave numbers 
smaller than 600 cm -1 , have been 
finished by L. Trafton. On this basis a 
preliminary "gray model" in radiative 
equilibrium has been constructed for 
Jupiter, which shows that without in- 
ternal sources 55 km-atm of H 2 provides 
sufficient thermal opacity to account for 
temperatures around 160°K at the visible 
cloud layer. 

Physics of Comets 

Claude Arpigny completed a critical 
discussion of some aspects of the physics 
of comets relating to their shape, the 
variation of radius with heliocentric 
distance, and the internal distribution of 
particles. He reviewed the abundance 
determinations in comets, with correc- 
tions and additions to the published 
literature, and completed a thesis en- 
titled "A study of molecular and physical 
processes in comets." On the physical 
side, his work embraced the velocities and 
accelerations of particles in the tails, the 
ionization of cometary molecules, and the 
different behavior of dust and gas tails. 

Dr. A. Stawikowski and Greenstein 
have studied the spectrum of comet 
Ikeya (1963a). This was a relatively 
dust-free bright comet which permitted 
very high resolution spectroscopy with 
the 200-inch telescope. Of particular 
interest is the presence of what seemed to 
be the isotope bands C 12 C 13 . The struc- 
tures of the (1,0), (2, 1) bands for normal 
and isotopic molecules were computed, 
as well as the available solar energy 
for resonance-fluorescent excitation. The 
strength of the total band head of the 
isotopic molecule was compared with 
that of a few individual rotational 
triplets of the normal molecule. On the 
basis of several different methods of 
analysis, an isotope ratio of C 12 /C 13 of 
about 70 with an error of about 20 per 
cent was deduced. This is so close to the 
terrestrial value that it seems probable 
that the carbon-isotope ratio is the same 



in cometary as in terrestrial and meteor- 
itic material. There is apparently very 
much less C 13 in the sun. A serious com- 
plication is the presence of several 
individual lines of NH 2 , one of which 
falls exactly on the C 13 C 13 band head and 
another on the C 12 C 13 head. From the 
expected relative intensity ratios it seems 
clear that most of the apparent C 13 C 13 
band in this comet is, in fact, NH 2 . If not, 
the abundance of C 13 deduced would be 
very high. However, since the C 12 C 13 
band is largely uncontaminated, the 

isotope ratio is fairly reliable. On the 
hypothesis of the origin of some of the 
rare isotopes by neutron bombardment in 
an early phase of the history of the solar 
system, this observation requires that 
the cometary material must once have 
been considerably closer to the sun than 
would have been expected. The cometary 
material would then come from approxi- 
mately the region of the terrestrial group 
of planets, rather than from the region of 
the major planets, if comets were formed 
at the same time as the solar system. 


Various Spectroscopic Investigations 

Deutsch has examined the available 
statistics of rotational line widths at 
various spectral types along the main 
sequence. He concludes that these statis- 
tics are fully consistent with a Boltzmann- 
type law for the distribution of angular 
momenta of stars — a situation that seems 
to have been widely misapprehended. 
One consequence of such a law is that the 
distribution law for V sin i must fall to 
zero with V sin i. With Kraft, Deutsch 
has begun to accumulate some line- 
width statistics for main-sequence F stars 
known to have V sin i < 25 km/sec. 

In collaboration with Dr. P. C. 
Keenan of the Perkins Observatory, 
Deutsch is continuing to study the spectra 
of selected Mira-type variable stars, with 
particular reference to the line weakening 
these stars show relative to non-Mira M- 
type giants. In some of these objects the 
line weakening undoubtedly signifies large 
metal deficiencies. He now finds a smaller 
line weakening to occur in some non- 
Mira M-type giants that have high 
velocities, but in most of these stars the 
lines are not perceptibly weakened rela- 
tive to those in non-Mira M giants of low 
velocity. Near the brightness maximum 
of February 1963, Mira itself exhibited a 
very peculiar spectrum, characteristic of 

so-called "weak-line cycles." The AlO 
bands were seen in emission. 

Deutsch is also continuing to observe 
circumstellar lines in a number of late- 
type giants and supergiants that are 
strategic objects for the futher elucidation 
of the mass-loss phenomenon that occurs 
in all such stars. In the profiles of the H 
and K lines, a number of stars have now 
been found to show marked variations in 
the parts (K 2 , K 3 ) that arise in the 
chromospheric and lower circumstellar 
levels. It was also found that in Septem- 
ber 1963 the M6 semiregular variable 
star CH Cygni showed a composite 
spectrum. A hot, blue continuum was 
superposed over the late-type spectrum, 
together with emission lines of H (strong 
and wide), He I (weak and wide), [Fe II] 
(strong and narrow), and Ca II (also 
strong and narrow). The spectrum of the 
recurrent nova T Coronae Borealis closely 
resembled this in June 1945, a few months 
before the nova outburst of 1946. No 
doubt there is a nova-like variable star 
in the CH Cyg system, too. However, a 
spectrogram of March 1961 showed no 
trace of the hot spectrum, which has 
faded appreciably since its discovery in 
September 1963. Radial velocity measures 
at 20 A/mm show no clear evidence 
during the past three years for orbital 
motion in this system. 



Greenstein and Oke have been making 
a search for field horizontal-branch stars. 
Both photographic spectra and photo- 
electric scans are being obtained. Of ten 
stars studied in detail, six have been 
found to be horizontal-branch-like ob- 
jects. Some of the remainder may be sub- 
dwarf stars, although they are rather 
hotter than would be expected. One 
star, BD-r-39°4926, has an effective tem- 
perature of 6800°K and an effective 
gravity of less than 10 cm sec -2 . It is 
likely that this star is similar to W 
Virginis stars but too hot to pulsate. 
There is, in fact, a little evidence that the 
star is slightly variable. 

Oke and Searle are making a survey of 
Ha, H/3, and H7 profiles in A-, F-, and 
G-type stars. For cooler stars, Ha is not 
affected seriously by absorption lines, 
and the profile should be usable for tem- 
perature determinations. Some of the 
computed profiles, however, do not match 
the observed profiles, possibly because of 
the uncertainty in the reasonance broad- 
ening term. This survey should indicate 
whether resonance broadening is pro- 
ducing the difficulty and will help de- 
termine how important it is. 

Baschek and Oke are making a study 
of A, Ap, and Am stars. For this purpose, 
high-dispersion spectra and photoelectric 
spectrum scans (X3400 to X10800) have 
been obtained. The scans have been 
corrected for absorption-line effects to 
produce continuum energy distributions. 
Using model atmospheres and hydrogen- 
line profiles computed by Mihalas, Bas- 
chek and Oke have found that the tem- 
peratures determined from the continua 
agree well with those obtained from the 
hydrogen lines. The B—V colors are 
seriously affected by line blanketing and 
cannot be used to infer temperature. 

At the 60-inch Newtonian spectro- 
graph, van Woerden has obtained spec- 
trograms at 85 A/mm for two-dimen- 
sional classification of some forty 0- and 
B-type stars in the region north of h and 
X Persei, where he had previously 

studied two neutral hydrogen complexes 
with abnormally high velocities. 

White Dwarfs 

Greenstein has continued his search 
for white dwarfs, particularly those 
having interesting spectroscopic features. 
The new proper-motion list from the 
Lowell Observatory has provided an 
excellent source for finding such objects, 
and approximately one-third of the 
Lowell proper-motion stars observed, 
which have yellowish or bluish color 
suspected, are in fact white dwarfs. 
Greenstein has also concentrated on 
spectroscopic properties of the cooler red 
degenerate stars and of white dwarfs in 
visual binaries. In the latter program, he 
has collaborated extensively with Olin 

Eggen and Greenstein have prepared a 
catalogue of some 150 white dwarfs with 
new photometric and spectroscopic re- 
sults. A color-luminosity relation (Mv, 
U—V) has been constructed for the stars 
with trigonometric parallax greater than 
0'. 05 (20 stars) and those that are mem- 
bers of binary systems and clusters (44 
stars). The moduli of 16 pairs in which 
the white dwarf is coupled with a K- or 
M-type dwarf have been obtained from 
(R, I) photometry, and the (Mv, R-I) 
relation has been obtained from nearby 
objects of large parallax. The (Mv, 
U—V) relation for white dwarfs is 
bifurcated with a richly populated, 
nearly horizontal branch among the 
bluer stars at M v = +ll m to +12 m , and 
a more steeply inclined group of fainter 
stars with a larger scatter in luminosity 
beginning at Mv near +12 m and con- 
tinuing to +15 m for the cooler stars. 

Within this group are found most of 
the white dwarfs with peculiar spectra, 
notably the yellow, carbon-rich white 
dwarfs of the X4670 type, the white 
dwarfs with continuous spectra, and 
the helium-rich white dwarfs of spectral 
type B. 

"White" dwarfs of quite red color have 



now been observed spectroscopically. 
Several of them should, in fact, be de- 
scribed as of spectral type DK ; they have 
weak lines of ionized calcium and, in the 
ultraviolet, blended resonance lines of 
neutral iron. It is now suspected for the 
first time that there are white dwarfs of 
spectral type DM. These stars have, in 
addition to features of DK type, a weak, 
shallow neutral-calcium line, X4227, a 
broad depression longward of this region, 
similar to the "pseudomolecular" de- 
pression in M dwarfs, and very weak 
TiO bands. The run of the equivalent 
widths of H7 in the normal DA-type 
white dwarfs shows a very interesting 
behavior; the equivalent width and the 
half-width of the lines increase to a very 
sharp maximum, then, within a few 
tenths of a magnitude in U—V color, the 
lines become sharper and, as they do so, 
disappear completely. The region of 
color index containing hydrogen lines is 
very wide and straddles the narrow region 
of color within which only helium lines 
are seen. It is clear that the surface 
composition of the white dwarfs is greatly 
variable, since hotter stars exist that have 
no helium lines. It is estimated that the 
helium-to-hydrogen ratio in a white 
dwarf showing helium lines is at least 
several hundred to one. No other features 
have as yet been found in the six known 
white dwarfs with broad, diffuse bands of 
molecular carbon. These must consist 
largely of carbon and helium. The exist- 
ence of some metallic lines in the cooler 
white dwarfs suggests that the complete 
gravitational separation of the metals 
from helium and hydrogen does not occur. 

Faint Blue Stars 

Greenstein has continued his observa- 
tions of the faint blue stars in the galactic 
polar regions. In collaboration with R. P. 
Fenkart he has discovered a new spectro- 
scopic binary, Tonantzintla 788, a pecu- 
liar subdwarf of type B. It has a velocity 
semiamplitude near 100 km/sec. 

The nuclei of the faint planetary 

nebulae discovered by Abell on the 
Palomar-Schmidt Sky Survey have been 
studied spectroscopically. They cover a 
very wide range of luminosity, from 
approximately to +7 visual absolute 
magnitude, as deduced from the spectral 
lines. It is interesting that a few of the 
hottest of these objects have lines of 
extraordinarily high excitation, notably 
O VI up to 128 volts excitation potential. 
It is also interesting that some of the 
most luminous of these nuclei, near 
absolute magnitude 0, are of the so- 
called Of type, with a combination of 
emission and absorption spectra. These 
Of stars are not of as high luminosity 
as those near the sun but represent 
some disturbed stage during the evolution 
of the star downward toward the white 
dwarfs. It is notable that in the Of 
planetary nuclei the lines present are 
those of carbon and oxygen; the usual 
nitrogen lines found in most high lumi- 
nosity Of stars are absent. It is very likely, 
therefore, that these stars belong to the 
carbon-rich sequence. 

An attempt to obtain the velocity dis- 
persion using prime-focus spectra of blue 
stars near the galactic pole is being made 
by Fenkart and Greenstein. The method 
of selection is such that only the W com- 
ponent of the space motion is obtained 
from the radial velocity, and, since W is 
seldom very large, one does not expect the 
highest-velocity groups to be present. 
Actually, they do not seem to be present; 
although the velocities are still prelimi- 
nary, it appears that the velocity dis- 
persion in the radial component will not 
be larger than 50 km/sec. From plates 
taken by Greenstein, Fenkart has ob- 
tained a velocity of —258 km/sec for 
the B2p star Barnard 29 in Messier 13, 
and —110 km/sec for the star 111-67 in 
Messier 13, a horizontal-branch A star. 

In collaboration with Dr. and Mrs. J. 
Berger, Zwicky obtained a number of 
plates on 103a-O and 103a-D emulsions 
taken with the 48-inch schmidt near the 
north galactic pole for study of the 



statistics of blue stars and for discovery 
of fast-moving p}^gmy stars. Triple- 
image plates were obtained for 10 fields 
around the north galactic pole and 25 
fields around the south galactic pole for 
the purpose of discovering Humason- 
Zwicky stars, blue stars in general, and 
blue compact galaxies. Radial velocities 
of a few dozen of the brighter blue stars 
are being measured on spectrograms 
obtained with the 60-inch telescope. The 
statistics of the blue stars in the Haro- 
Luyten Catalogue have been studied, 
using Zwicky's dispersion-subdivision 
method. The main results are that the 
blue stars do not form any decided 
swarms or clusters and that they extend 
far out into the halo of the Galaxy. 

Absolute Spectrophotometric Standards 

A new set of 13 absolute spectrophoto- 
metric standard stars in the declination 
range ±15° have been observed inten- 
sively, and absolute energy distributions 
have been obtained by Oke. All have been 
observed in the blue and more than half 
in the infrared. Comparison with pre- 
vious calibrations of these stars and with 
observations by Dr. K. Banner at Mc- 
Donald Observatory indicate that system- 
atic errors are no more than 0.01 to 0.02 
magnitude over the spectral range X3400 
to X10800. Based on all available data 
on the absolute calibration of a Lyrae, a 
new absolute energy distribution of a 
Lyr has been adopted. It corresponds to 
an effective temperature of 9500°K. 

Oke has obtained photoelectric spec- 
trum scans of a selection of Hyades stars 
down to spectral type G5. The scans 
cover the region from X3400 to X10800. 
Conti and Oke are using slit spectra to 
correct the scan observations to the con- 
tinuum by removing the effects of ab- 
sorption lines. This is being done chiefly 
for wavelengths greater than 5000 A. 
From the final absolute energy distribu- 
tions of the continuum, effective tempera- 
tures will be determined by comparison 
with model atmospheres. 

Approximately 60 main-sequence stars 
with spectral types between G6 and M2 
are being observed by Whiteoak with the 
Cassegrain photoelectric spectrum scan- 
ner mounted on the 60- and 100-inch 
telescopes. The scanner is used as a 
monochromator with a passband of 50 A. 
Intensities of each star's continuum are 
obtained at 18 regions free from line 
absorption between X5000 and X10800. 
For each star, the scanner observations 
are reduced by an IBM 7094 to a series of 
relative magnitudes with the zero point 
at X5560. The spectral responses of the 
telescope mirrors, the scanner, and the 
RCA 7102 photomultiplier are eliminated 
by the observation of standards for which 
Oke has measured absolute energies. 
Since the atmospheric extinction con- 
stants are small within the spectral range 
adopted, mean values are used. Pre- 
liminary reductions suggest that the 
observed stars, which were selected from 
a list studied spectroscopically by Wilson 
a few years ago, show the same variation 
of continuum temperature within each 
spectral type as inferred from Eggen's 
P—V colors. 

Color Systems 

Dr. Fenkart has studied methods of 
calibrating the Becker R, G, U system, 
used extensively for photographic sur- 
veys. Scans over different sets of standard 
wavelength between X3000 and X10000 
for 20 Hyades stars were reduced to a 
consistent scale together with a Lyr and 
10 Lacertae. Nine of these stars, with 
standard wavelengths chosen accurately 
enough to allow for the line features, 
were integrated at different air masses 
and different reddenings over the colors 
of the U, B, V and the R, G, U systems 
to establish semitheoretical main se- 
quences in both systems. In addition, this 
permits the study of the reliability of the 
calculated relationship between the two 
systems of Becker and of Johnson, and 
also the slope of the reddening paths 
adopted in the R, G, U system. Fenkart 
is also examining the high-latitude region 



near Selected Area 54, on the R, G, U 

Chemical Composition of Stellar 

Conti has concentrated on the study of 
abundances of the chemical elements in 
the A and F stars. He has completed a 
study of the abundances of five metallic- 
line A stars in the Hyades relative to 
normal F stars in the cluster, using semi- 
empirical model atmospheres. No model 
could explain the anomalous line 
strengths in the Am star, for elements 
such as calcium and nickel, so that ap- 
parently the abundance differences are 
real. He is completing work on the main- 
sequence stars in the Hyades, types A5 to 
K2. Preliminary results are that all stars 
have normal element-to-iron ratios, al- 
though at present there is some un- 
certainty in the iron-to-hydrogen ratios 
because of uncertainties in model at- 
mospheres. His goal was to study the 
effect of blanketing along the main 
sequence, and any possible convection, 
on the temperature as a function of 
optical depth. He has found two A 
stars, one in the Hyades, one in Praesepe, 
which are above the normal cluster turn- 
off points. The one in the Hyades is a 
metallic-line star; that in Praesepe seems 
to be normal in every respect. Peculiarity 
among metallic-line stars near the main 
sequence is so widespread that almost all 
sharp-line stars falling between +0.05 and 
+0.35 magnitude, in B—V color, are 
metallic-line objects. In the Pleiades, 
however, there are no metallie-line A 
stars yet known, and Conti will study 
observational material on A0 and F0 
stars there. 

A study (in collaboration with Waller- 
stein) of the K giant stars in several inter- 
mediate-age galactic clusters has not 
shown any peculiar abundance anomalies 
among various metals, although two of 
the clusters more closely resemble the 
Hyades abundances and two others 
those in the sun. 

The study of the spallation-product ele- 

ment lithium is continuing. Conti has 
found lithium in the very old subgiant 
8 Eridani; this star has at least twice as 
much as the sun. Because of the con- 
vective mixing in a K subgiant, it is 
almost certain that 8 Eri has produced 
lithium during its main-sequence history 
because of the more rapid destruction 
rate at present. Danziger and Conti 
have found a number of sharp-line main- 
sequence early F stars in which the 
element is present. They offer some 
opportunity for the determination of the 
lithium to beryllium ratio in stars with 
enhanced lithium. The difficult observa- 
tions of the beryllium lines in the far 
ultraviolet are being initiated. 

Gaustad has obtained spectra of a 
few of the B stars in the Orionis associa- 
tion which have anomalously weak 
helium lines for their spectral type. 

Conti and Danziger have obtained 
spectra of 10 sharp-line B stars to 
determine abundances and the possible 
effects of rotation on the atmospheres of 
similar objects. The sharp-line star in the 
Pleiades, Maia, is included in this list. 
Danziger is studying the hot subdwarf 
stars such as HD 137569, computing 
detailed model atmospheres, to deter- 
mine the helium and metal content of the 
stars. He has also obtained a preliminary 
model atmosphere for HZ 44 to analyze 
lines measured in this hot subdwarf by 
Greenstein and Munch. The slightly 
weak-line F star, a Bootis, which is sus- 
pected of being subluminous, is being 
compared with the other nonrotating F 
stars to obtain abundances relevant to 
those in the Hyades. Other effects of 
rotation on the spectra of F stars are 
being sought by Danziger. 

Wyller has attempted to clarify the 
problem of the C 12 /C 13 abundance ratio 
in the carbon-rich stars. There has been 
an anomaly in the sense that the C 13 -rich 
stars, as indicated by the C 2 bands, were 
not confirmed by the earlier search by 
Wyller on Mount Wilson high-resolution 
plates for the (2, 0) C 13 N 14 bands. To 
clarify the problem, since from Bouigue's 



work the molecular constants were ap- 
parently insufficient to predict accurately 
the location of individual C 13 N 14 rota- 
tional lines, Wyller constructed a dis- 
charge tube using an electrodeless dis- 
charge filled with C 13 -enriched methane, 
nitrogen, and argon gases. The experi- 
ment was successful, and he has obtained 
excellent spectra of the infrared CN 
bands. On the basis of the laboratory 
wavelengths now obtained, individual 
C 13 N 14 rotational lines were identified in 
the spectra of several early- and late- 
type carbon stars. The provisional C 12 / 
C 13 abundance estimate was made from 
the relative intensities of selected normal 
and isotopic rotational lines in the star Y 
Canum Venaticorum. C 13 appears to be 
strong, and the abundance ratio C 12 /C 13 
is somewhere between 2 and 3. 

Stawikowski is investigating in detail 
abundances in several selected metallic- 
line stars of various degrees of metal- 
licism. He is using high-resolution plates, 
spectral scans corrected for line blanket- 
ing, and model atmospheres. 

Dwarf stars of spectral type K5 and 
later have been observed photoelectri- 
cally with the Cassegrain scanner to 
determine the strength of their molecular 
features and the possible use of molecules 
for differentiating between late-type 
dwarfs of population I and halo popula- 
tion II. Intensity measurements taken in 
the X6380 band of CaH, the X7054 band 
of TiO, and the continuum region of 
X7450 in these late-type dwarfs have 
been used to separate the two popula- 
tions. The separation depends on measure- 
ments of the metal deficiency of the 
population II stars, which could be 
evident in the difference in the band 
strength ratio of CaH/TiO when dwarfs 
of similar spectral class but different 
populations are compared. From the 
ratio of the band strengths of the TiO 
molecule in stars of the two populations, 
an estimate of the difference in effective 
temperature of stars with the same 
spectral class but different abundances 
can be made with the help of a simple 

model atmosphere. Values obtained are 
then compared with observationally 
measured colors taken over a broad 
spectral range covering the visible and 
infrared regions of the spectrum. This 
possible differentiation of the two types 
of M dwarfs is of particular significance, 
since the normal photographic region of 
the spectrum does not show very great 
differences in the spectra of high- and 
low- velocity stars. 

Colors and Magnitudes 

Eggen has obtained colors and magni- 
tudes on the U, B, V system for 228 
visual binaries for which orbits are avail- 
able. Also, photometry of additional 
physical companions to 26 systems has 
been obtained with the 100- and 200-inch 
reflectors. The components of most of 
these orbital systems differ by less than 
0.1 or 0.2 magnitude. The mean masses 
are discussed. For several systems con- 
taining K- or M-type dwarfs, and for 
late-type, common proper-motion com- 
panions to several other systems, V, R, I 
photometry has been obtained with the 
60- and 100-inch reflectors and an RCA 
7102 photomultiplier. A calibration of the 
(Mv, R — I) relation, obtained from stars 
with large trigonometric parallax, yields 
good photometric parallaxes for these 
systems. New orbits have been computed 
for 43 systems (ADS 148, 713, 864, 999, 
1833, 2630, 3520, 4153, 4971, 5949, 7131, 
8166, 8337, 8344, 8635, 8680, 8954, 9094, 
9185, 9264, 9380, 9397, 9441, 9689, 10158, 
10871, 11842, 12911, 12961, 14412, 14424, 
15267, 16436, 16836, 17178, HD 2885, 
HD 20121, HD 27019, HD 104747, HD 
194433, HD 206644, 1 Geminorum, and 
+ 27°2853) with an IBM 7090 using a 
program written by John Castor. The 
presence of two discrete M-L relations is 
confirmed. One (called the Hyades M-L 
relation), which is valid for the Hyades- 
Pleiades stars, is linear in the (M B , log m) 
plane with a slope of L ~ m 2 from Mb = 
— 2 m to +10 m ; the other (called the sun- 
Sirius relation), valid for the remaining 
objects, contains no stars brighter than 



about M v = +2 m and has a slope of 
L ~ ra 3 . Neither of these slopes agrees 
with the relation L ~ ra 4 - 8 , obtained from 
the available stellar model computations. 
The stars falling on the Hyades M-L 
relation are apparently all as young as or 
younger than the Hyades cluster and 
have galactic orbits similar to those of 
the Hyades and Pleiades clusters; these 
objects have never been outside a galactic 
belt about 2 kpc wide. The objects falling 
on the sun-Sirius M-L relation are a 
mixture of objects older than the Hyades, 
and they have galactic orbits that carry 
them outside the 2-kpc-wide belt popu- 
lated by the Hyades-Pleiades stars. 

Eggen has obtained colors and magni- 
tudes for about 200 randomly selected 
stars between V E = 8 m and 18 m near the 
north galactic pole (SA 56, SA 57, and 
region of Coma Berenices cluster). Some 
150 of these have values of B — V be- 
tween +0?3 and +0™8, and values of 
the ultraviolet excess, 8(U — B), can be 
derived. A minimum value of Z for each 
star has been obtained by correcting the 
observed B— V for blanketing effects and 
fitting to the Hyades main sequence. 
Values of 8(U-B) from -0 m 03 to 
+0?28 are found for stars with Z < 400 
parsecs; values from +0 I ?03 to +0 m 28, 
for 400 < Z < 800 parsecs ; and values of 
+0 m 15 to +0 m 28, for Z > 800 parsecs. 
The lack of stars with 8(U—B) less than 
+0^15 at values of Z > 800 parsecs is 
consistent with the previous suggestion 
by Eggen, Lynden-Bell, and Sandage that 
the first generation of stars was formed 
during a rapid collapse of the galaxy and 
that subsequent generations have been 
formed in the resulting disk, which is less 
than 2000 parsecs thick. 

Eggen has made five-color observations 
(U, B, V, R, I) of some 300 of the K- and 
M-type dwarfs brighter than 9.8 visual 
magnitude in the lists published by Vys- 
sotsky and his colleagues. Proper motions 
and radial velocities (from Wilson's 200- 
inch coude plates) are being collected 
for the determination of space motions. 
Three-color observations (U, B, V) of 100 

stars in the region of 5 Lyrae to V E = 18 m 
have been made. The existence of a 
cluster, suggested by Stephenson, is 
confirmed. The main sequence extends 
from V E = 5 m to 14 m and resembles the 
Pleiades; in fact, the space motion of 
the cluster is nearly identical with that of 
the Pleiades. The modulus is 7 m 6, and 
there is little or no absorption in the 
region. The resulting luminosity of 8 
Lyrae (M4 II) is -3 m 3. 

Eggen and Thomas Greenfield obtained 
2000 observations with the 20-inch re- 
flector of K- and M-type giants brighter 
than visual magnitude 5.5. 

Emission Lines of Ionized Calcium 

Work reported last year by O. C. 
Wilson indicated that the chromospheric 
activity of a main-sequence star is proba- 
bly a decreasing function of time. This 
tentative conclusion was based on a 
spectroscopic study of the intensities of 
H and K emissions in the components of 
visual binary systems, in single field 
stars, and in the members of the four 
clusters Hyades, Praesepe, Pleiades, and 
Coma. The one large uncertainty re- 
maining was that chromospheric prop- 
erties of stars formed in large clusters 
might differ significantly from those now 
located in the general field. This point 
has now been settled by a joint investiga- 
tion with Dr. A. Skumanich of the High 
Altitude Observatory, which makes use 
of unpublished multicolor photometry 
by Stromgren and Perry. 

Two of the photometric indices of 
Stromgren and Perry, Ci and (b — y), are 
so chosen that c\ is a measure of lumi- 
nosity and (b — y) a measure of color, in- 
sensitive to other parameters. When the 
stars are plotted in a c h (b — y) diagram, 
the main sequence appears as a band of 
points whose lower edge represents the 
zero-age, essentially unevolved stars. 
Spectroscopic observations of 142 of these 
objects have been obtained at 10 A/mm 
dispersion, beginning at about spectral 
type F7 and extending to about type G2. 
After elimination of known spectroscopic 



binaries and other stars whose spectra 
show that they do not belong to the main 
sequence, a sample of 114 main-sequence 
stars remains. Among them, Wilson finds 
17 stars with bright H and K, of which 
15 lie along the zero-age edge of the 
distribution. Assuming that the 2 excep- 
tions that occur higher in the diagram 
are probably undiscovered close binaries, 
the evidence is very strong that chromo- 
spheric activity does, indeed, decrease 
with age in main-sequence stars. 

The foregoing result, together with 
those reported earlier, suggests that the 
intensities of bright H and K lines could 
be used to determine the relative ages 
of groups of stars and thus provide infor- 
mation on the rate of star formation 
during past epochs. For this purpose, a 
sizable sample of late-type main-sequence 
stars is required, and it must not have 
been chosen with reference to any kine- 
matic properties but solely on the basis of 
spectra. Fortunately, such a sample is 
known, thanks to the efforts of A. N. 
Vyssotsky and his colleagues. Therefore, 
observations have been started by Wilson 
with the 8-inch camera of the 200-inch 
coude spectrograph, covering a large 
fraction of the stars in Vyssotsky 's 
catalogue. Besides being used to derive 
H-K emission intensities, the spectro- 
grams are being measured for radial 
velocity; they also provide spectral 
types. A preliminary and partial analysis 
of the first 96 stars observed has been 
published, and the results appear suffi- 
ciently promising to continue the work. 

All the measures of widths of H and K 
emission have been collected by Wilson, 
and it is hoped to publish this material, 
together with the resulting color-magni- 
tude diagram for field stars, in the near 

Absolute Magnitudes from the Width of Ha 

Together with George Preston of Lick 
Observatory and Sidney Wolff of the 
Berkeley Department of Astronomy, 
Kraft has shown that the width H of the 
core of Ha in absorption in the spectra of 

late-type stars increases with increasing 
luminosity. Among G- and K-type stars, 
a correlation, independent of spectral 
types, exists between log Ho and the 
absolute ultraviolet magnitude Mu> It 
was found that M-type giants could not 
be distinguished from supergiants, but 
dwarfs could easily be differentiated from 
giants. Unfortunately, the relationship 
fails for class IV stars of all spectral types, 
because of the influence of damping wings. 
Emission was detected as a probable 
distorting agent in the profiles of Ha 
among many supergiants of type K2 and 
later. The core of Ha arises in the chromo- 
sphere and is probably Doppler in origin; 
its dependence on absolute magnitude is 
seen as an analogue of the Wilson-Bappu 
effect in the K2 emission of Ca II. 

The main application of the foregoing 
relationship as a tool for distance deter- 
mination may be found in galactic- 
structure studies because the Ha measure- 
ment may conceivably be extended to 
supergiants of apparent magnitude about 
+ 10, whereas the K2 method, which 
requires high-dispersion spectrograms 
greatly overexposed (for the continuum), 
is confined to stars with m v ^ +7. On the 
other hand, the Ha method is probably 
less accurate than K2 in practice. Under 
current investigation is the possibility 
of using the equivalent width, rather than 
the half-width, of Ha. 

Kraft and 0. C. Wilson have esti- 
mated the strength of Li I (X6708) against 
X6703 of Fe I and X6718 of Ca I in the 
spectra of about 30 main-sequence F- 
and G-type stars. The stars were selected 
from Stromgren and Perry's photometric 
catalogue on the b — y, Ci, mi system in the 
interval 0.33 < b-y < 0.42, without re- 
gard to Ci or m h or Ca II emission 
strength. From the limited material at 
hand, the following conclusions are 
reached : 

1. All stars showing strong Ca II 
emission show strong Li I in absorption, 
but Li I absorption occurs in strength in 
many stars without Ca II emission. 

2. If, following Stromgren, we inter- 



pret the index c\ as a nuclear age pa- 
rameter, there is no correlation between 
Li I strength and nuclear age. 

3. There is no correlation between Li I 
strength and the metal index mi. 

A few observations of main-sequence 
binary systems have been made by 
Wilson to test for the presence of Li I. 

Influence of Rotation on Colors and 

Magnitudes of Main-Sequence 

F-Type Stars 

Kraft and Marshal Wrubel of Indiana 
University have shown empirically that, 
at a given B — V, the slow apparent 
rotators (v sin i small) of the Hyades 
show an ultraviolet excess [8(U — B) > 0] 
relative to the fast rotators for stars 
with spectral types between A8 V and 
F5 V. The effect amounts to about 0.06 
magnitude in B — V for the extreme cases 
(v sin i ~ 200 km/sec); metallic-line 
stars have been excluded. 

The order of magnitude of the effect is 
fully predicted by the change in surface 
gravity and effective temperature com- 
puted by Sweet and Roy ; its empiri- 
cal calibration in the U—B versus B—V 
diagram is based on work by Gunn and 
Kraft. The theory shows that slow 
rotators and rapid rotators seen pole-on 
cannot be distinguished by 5(U — B), in 
accord with observation, but they might 
be distinguished in a color- magnitude 
array. Unfortunately, the back-to-front 
magnitude difference in the Hyades 
precludes the possibility of carrying out 
the latter test. Further work is planned 
on Praesepe. 

The principal result of the investiga- 
tion seems to be that mild ultraviolet 
excesses for field F-type stars cannot be 
interpreted directly as evidence of 
metal deficiency unless the rotation is 
average to large for that spectral type. 

U Geminorum Stars 

A new, more nearly definitive, orbit has 
been obtained by Kraft for Z Camelo- 
pardalis, and the spectroscopic orbital 
period of P = 0.28066 day has been im- 

proved to P = 0.2806634 day by means of 
photoelectric observations by W. Krze- 
minski. Z Cam has grazing eclipses, and 
the masses of the blue and red com- 
ponents appear to be 1.1 and 1.3 solar 
masses, respectively. The mean error of 
these determinations is large, probably 
around ±0.2 solar mass, principally be- 
cause the absorption lines of the red star 
are extremely evanescent. Despite its 
crudity, the determination is nevertheless 
significant since the object is the first of 
its kind for which a direct estimate of 
mass has been obtainable. 

Kraft has collaborated with W. J. 
Luyten of the University of Minnesota in 
making a new estimate of the mean abso- 
lute magnitude of U Gem stars at mini- 
mum light, based on 15 proper motions 
and 10 radial velocities. Omitting EX 
Hydrae,which is a high-velocity object,the 
solar motion is found to be +43 ± 15 km/ 
sec, and the apex is near A = 18 h , D = 17°. 
From t components of the proper motion, 
we get < Mv >min ^ +8.2, a value quite 
insensitive to reasonable changes in the 
solar motion. It appears that the stars are 
about 10 times too faint for the apparent 
spectroscopic luminosities of the red 

Old Novae 

New spectrograms obtained by Kraft 
at the prime focus of the Hale reflector 
show that Nova Puppis (1942) has large 
radial velocity variations in a short time 
and is presumably a spectroscopic binary. 

Kraft and W. Krzeminski of Warsaw 
University have advanced a new model 
for WZ Sagittae (1913, 1946) based on 
3000 ultraviolet photoelectric observa- 
tions and simultaneous photoelectric and 
spectroscopic runs over 2 x /i periods (P = 
0.05668786 day). It is established that 
eclipse curves of the W Ursae Majoris 
type and the radial velocities from the 
hydrogen emission lines are approxi- 
mately 90° out of phase. Maximum 
velocity of recession occurs near principal 
minimum. The new model attributes this 
effect to a rapidly receding stream ejected 



from the dark star toward the primary 
component. The properties of the radia- 
tion field of the white-dwarf primary are 
derived from tracings of H7 in absorption 
and are shown to be consistent with the 
observed geometrical extent of the line- 
emitting regions. The best present esti- 
mates of the geometrical and dynamical 
parameters of the system are: Sflfli = 
0.59 SfTCo, 3H 2 = 0.03 9Tlo, STIs/STl! = 0.05, 
i ~ 82°, R x = 0.87 X 10 9 cm, R 2 ~ 7 X 
10 9 cm, a = 3.7 X 10 10 cm, where the 
subscript 1 refers to the white-dwarf pri- 
mary. The numbers are rather uncertain, 
but there is little doubt that the mass 
ratio is unusually small. The proposed 
emission of gravitational waves (Kraft, 
Matthews, and Greenstein, 1962) is 
such that the gain in phase is 1 second in 
12 years. It is likely that mass transfer 
and mass loss during outbursts of WZ 
Sge severely mask the period change 
resulting from the emission of gravita- 
tional waves. 

T Tauri Stars 

The spherically symmetric model for 
mass loss discussed previously by Kuhi 
(dissertation, Berkeley, 1963) was used in 
analyzing further spectrograms of T Tau 
stars supplied by Greenstein. The results 
were in essential agreement with those 
obtained earlier (Publ. Astron. Soc. 
Pacific, 75, 415, 1963) for stars also 
observed at Lick Observatory (T Tau, 
RY Tau, GW Orionis). Three additional 
stars were also considered, for which the 
preliminary values for the rates of mass 
loss are as follows: 

RW Aurigae: 1.1 X 10 -7 solar mass per year 
SU Aurigae: 2.5 X lO" 8 

R Monocerotis: 7.8 X lO" 8 

However, the larger number of spectro- 
grams available makes it clear that the 
model used cannot adequately represent 
the mass ejection for all the T Tau stars 
studied, in particular T Tau itself, which 
has consistently narrower emission lines 
than predicted. 

Kuhi carried out a preliminary analysis 

of the effect of mass ejection on the time 
scale of contraction by making use of 
Iben's theoretical tracks. These tracks 
provide L,R,dR, and dR/dt as a function 
of time from some initial configuration 
for a contracting star in convective 
equilibrium. An additional term repre- 
senting only the kinetic energy of the 
mass loss was then included. Two possible 
cases were considered: (1) mass ejection 
most violent initially but gradually de- 
creasing as the star approaches the main 
sequence; (2) mass ejection at some 
average rate throughout its contractive 
life. The rates for both cases were de- 
termined from the observed profiles and 
the distribution of stars among the differ- 
ent emission-line intensity classes. For a 
1.5-solar-mass star, the overall effect on 
the time scale seems negligible, i.e., about 
a 1 per cent decrease in contraction time 
for case 1 and 2 per cent in case 2. How- 
ever, the mass loss is a substantial frac- 
tion of the total — about 0.17 solar mass — 
and in the earlier stages the kinetic 
energy involved is about 16 per cent of 
the radiative luminosity. This clearly im- 
plies that energy transfer by means of 
mass ejection may have to be considered 
in the construction of pre-main-sequence 
evolutionary models. It also indicates that 
the analysis attempted is inadequate, 
because such large changes must affect 
the internal structure of the star, and 
that a better treatment is necessary. 

Discrete spectral scans of AS 209, a T 
Tau star with bright ultraviolet, have 
been obtained to determine its spectral 
energy distribution so that the origin 
of the excess ultraviolet can be discussed 
more critically. 

Wolf-Rayet Stars 

Observations of binary and single 
Wolf-Rayet stars brighter than tenth 
magnitude are under way by Kuhi with 
the Cassegrain photoelectric scanner used 
essentially as a narrow-band photometer. 
The intensities of various emission lines 
of different ionization and excitation are 
being measured in the wavelength range 



X3200 to XI 1500, not only to provide more 
reliable intensities wherever possible but 
also to detect possible intensity variations 
and anomalies. These variations are of 
greatest interest in the three eclipsing 
binaries CV Serpentis, CQ Cephei, and 
V444 Cygni, where such variation with 
phase can be expected to give some clues 
to the structure of the Wolf-Rayet atmos- 
phere. Observations in the fall of 1963 
confirmed Hiltner's earlier results for the 
behavior of He II X4686 ; that is, it reaches 
maximum intensity at each conjunction 
and is also intrinsically variable. Pre- 
liminary results for V444 Cyg seem to 
indicate that lines of higher ionization 
undergo a narrower eclipse than those of 
lower ionization, whereas the He II X4686 
changes very little. This implies that the 
ionization decreases outward in the 
atmosphere and that the He II emission 
occurs in a very large envelope surround- 
ing the star. 

The star HD 50896 found by Wilson to 
have irregular radial velocity variations 
has also been found by Kuhi to show 
irregular variations in emission-line in- 
tensity; it is as much as 0.20 magnitude 
for He II X5412, and usually about 0.10 
magnitude for most other lines, including 
He II X4686, N III X4633, and N IV 
X4057. Most emission lines show some 
such fluctuation and thus seem to be indic- 
ative of irregular variations in the Wolf- 
Rayet atmosphere itself. 

Observations of binaries with different 
periods are being made in order to detect 
possible interaction effects between the 
W and O-B components. 

Galactic Clusters 

The old galactic cluster NGC 188 has 
been studied by Greenstein and Keenan 
on prime-focus spectra obtained with the 
Hale telescope. The brightest stars are of 
spectral type Kl and seem to be essen- 
tially normal giants with, if anything, 
slightly strong cyanogen. There are no 
indications of metal deficiency. There is 
some color excess, indicated by a com- 
parison of the available photoelectric 

colors with the spectral types. This color 
excess is larger than that indicated by the 
photoelectric calibrations available; it 
would suggest that the cluster diagram of 
NGC 188 should be shifted somewhat 
toward that of M 67 and that the age of 
NGC 188 should be reduced. The radial 
velocities of 11 stars were measured, 
yielding a mean cluster velocity of 
— 49 km/sec. Most of the bright stars 
seem to be members, in so far as the 
radial velocity provides a discriminant. 

New three-color photometry of about 
75 stars in NGC 188 has been completed 
by Eggen and Sandage. The older photom- 
etry of Sandage is confirmed in V and 
B—V, but a zero-point error of 0^04 
in. U — B was found. Analysis of the new 
data along lines similar to those reported 
for M 67 last year shows E(B - V) = m 10 
and d(U-B) = +0 m 03. This reddening 
value is now in satisfactory agreement 
with Abt and Golson's reddening value 
for field stars in the neighborhood of 
NGC 188 and removes the discrepancy 
noted in the first study of NGC 188, 
which gave E (B -V) = +0 m 05. The data 
show a gap in the evolving main sequence 
of NGC 188 which occurs m 5 fainter 
than the main-sequence termination 
point. It is similar to the gap established 
last year in M 67 and arises when the 
hydrogen content goes to zero at the 
center of the evolving stars, causing 
gravitational contraction and a hydro- 
gen-burning shell source to be established 
at the edge of the helium core. The dis- 
tance modulus of NGC 188 based on these 
new data is (m — M)o = 10.90. The main- 
sequence termination luminosity is 
M Vq = +3.85, which keeps NGC 188 as 
the oldest disk population cluster known. 
Comparison of NGC 188 with the new 
data on M 67 reported last year shows 
that the main-sequence termination lumi- 
nosities differ by AM v c^ 0™70 rather than 
AM, ^ l m obtained several years ago. 

A composite color-magnitude diagram 
showing M 67, NGC 188, and the five 
globular clusters completed this year 
indicates that the main sequence of the 



globular clusters terminates halfway be- 
tween the breakoff points of M 67 and 
NGC 188. However, the age of the 
globular clusters can be made longer 
than that of NGC 188 by a slight differ- 
ence in the hydrogen abundance between 
disk and halo clusters in the sense 

^globular cluster > X NGC 188- 

Photographic plates and photoelectric 
sequences in the galactic clusters M 37 
and NGC 2420, of intermediate and old 
age, respectively, were turned over to 
Fred West at the University of Indiana. 
The results of the analysis of this ma- 
terial have just become available as his 
Ph.D. thesis. 

Further work on light curves of long- 
period variables in the globular cluster <a 
Centauri has been done by Arp, Brueckel, 
and Lourens. Another small-amplitude, 
41 -day period was found at the end of the 
giant branch in co Cen. There are other 
similarities to the behavior of long- 
period variables in 47 Tucanae which 
may be illuminated when the two clusters 
are compared more closely. Variable 2 
appears to have a period of 230 days 
with a variable maximum (by about 1 
magnitude) and a total amplitude be- 
tween 5 and 6 magnitudes. Unlike the 
200-day variables in 47 Tuc, variable 2 
seems to fall considerably fainter than 
the end of the giant branch. 

Globular-Cluster Photometry 

Three-color photoelectric photometry 
of the clusters M 3, M 13, M 15, and M 92 
has now been completed and analyzed by 
Sandage and Katem after four years of 
observation. Photoelectric data are now 
available for 53 stars in M 3 between 
V = 12 m 7 and V = 21 m 0, for 84 stars in 
M 13 between V = 12?0 and V = 21 m 8, 
for 51 stars in M 15 between V = 12 m 8 
and V = 21 m 8, and for 73 stars in M 92 
between V = 12 m and V = 22 m l. Dis- 
tance moduli were obtained for all four 
clusters using the full machinery of the 
method of photometric parallaxes with 
reddening, blanketing, and evolutionary 
corrections applied. Reddening corrections 

were found by the following three meth- 
ods: (1) comparison of the observed B — V 
colors of the main-sequence termination 
point in each cluster with the bluest 
B—V color of the nearby field sub- 
dwarfs of extreme metal weakening as 
reported last year ; (2) three-color photom- 
etry of many field stars in the direction 
of each cluster together with the proper- 
ties of the reddening trajectory in the 
U — B, B—V diagram; and (3) com- 
parison of the U — B, B — V diagram for 
the bluest stars on the horizontal branches 
of each cluster with an intrinsic line in 
this diagram calibrated by stars in M 92 
whose accurate E{B—V) is known from 
methods 1 and 2. Blanketing corrections 
were determined from the index of line 
weakening 8(U — B) by measurements of 
giants and dwarfs in each cluster. The 
evolutionary corrections were obtained 
from calculated models by Hoyle. 

When the data were corrected for 
reddening and blanketing, photometric 
fits to the Hyades main sequence were 
made using the globular-cluster main- 
sequence data following the precepts of 
Eggen and Sandage in the Astrophysical 
Journal, volume 136, page 735, 1963. 

The resulting distance moduli directly 
give the absolute magnitude M v of the 
RR Lyrae stars in each cluster with the 
result that M v = +0 m 42 ± 0.08 (M. E.), 
excluding the anomalous case of M 13 but 
including data for 47 Tuc from the pho- 
tometry of W. Tifft. The data for the 
individual clusters are shown in table 1. 

The observed ultraviolet excess values 
8(U—B) correlate very well with the 
metal weakening of Fraunhofer lines as 
determined spectroscopically by Morgan 
and by Deutsch. There appears to be no 
correlation of M v (RR Lyr) with metal 
weakness, and it can be concluded that 
the horizontal branches of globular 
clusters fall at the same M v regardless of 
the metal abundance, at least for the 
five clusters studied here and for M 5 
studied previously by Arp. 

The < M v > for the two RR Lyr stars 
in M 13 is brighter than for the other 








M v (RR Lyr) 


m 00 

m 15 



M 13 





M 15 





M 92 







(0 . 10) 

13 . 40 


clusters. Inspection of the color-magni- 
tude diagram suggests that the variable 
stars in M 13 may come along evolu- 
tionary tracks associated with the asymp- 
totic branch near the giant branch rather 
than from the horizontal branch as in 
the other clusters. 

Comparison of the main sequences of 
all five clusters fitted together according 
to their (m— M) values shows that the 
main-sequence termination luminosities 
are identical to within the accuracy of 
the data (±0 m 2). Therefore, the present 
data suggest that the ages of the globular 
clusters in the galactic halo are equal to 
within ±20 per cent— a significant re- 
sult for unraveling the early history of the 
galactic system. 

Absolute ages cannot yet be deter- 
mined because of the severe dependence 
of age on the abundance of hydrogen and 
helium. However, adopting Hoyle's cal- 
culated evolutionary model, the present 
data suggest that T = 18 X 10 9 years if 
X = 1.00 and! 7 = 12 X 10 9 years if X = 
0.80. The lower ages are currently 

favored on the basis of determination of 
X from (1) the planetary nebula in M 15, 
(2) Traving's chemical analysis of the star 
Barnard 29 in M 3, and (3) evidence 
from the position of the main sequence 
in the M b i, log T e plane for subdwarfs 
compared with population I stars. 

Properties of the Galactic Center 

Arp has submitted for publication a 
paper that discusses extensive observa- 
tions in the direction of the Milky Way 
center (window centered on the globular 
cluster NGC 6522). It is concluded that 
these central stars are essentially only 
metal-rich stars ranging from moderately 
rich (metal-rich globular cluster) to very 
metal rich (solar-neighborhood-type star) . 
Most of the stars are older than 10 9 years. 

It is shown that the nucleus of our 
Galaxy is only about half the size of the 
nucleus of the Andromeda nebula, and 
consequently that our own Galaxy should 
be classified as an Sc rather than an Sb 


Combination of Optical and Radio Data 

With the 100-inch coude spectrograph, 
van Woerden has continued his high- 
resolution studies of interstellar lines in 
the spectra of early-type stars. The 
purpose of this work is a better under- 
standing of the structural and kinematical 
properties of interstellar clouds, to be 
gained from a combination of spectro- 
graph^ data on the Ca II absorption lines 

in the violet and radioastronomical 
information on the neutral hydrogen 
emission line at 21-cm wavelength. The 
extensive hydrogen-line profile measure- 
ments made at Dwingeloo, the Nether- 
lands, offer a velocity resolution of 2 
km/sec. The most detailed Ca II observa- 
tions published so far are those of Adams 
(1941-1949), with a dispersion of 2.9 
A/mm and about 9 km/sec resolution. 
Van Woerden has made intensive use of 



the higher dispersions now available, 1.1 
and 1.9 A/mm (to be called a and b), 
which give velocity resolutions of the 
same order as that in the hydrogen line; 
a dispersion of 4.4 A/mm (c) was em- 
ployed for fainter stars, mostly unob- 
served by Adams. In the two years of 
his fellowship, van Woerden obtained a 
total of about 250 spectrograms for 150 
stars. The distribution over different 
areas of sky is as follows. 

In the Orion region, 20 stars were ob- 
served at dispersion a, 19 stars at b, and 
21 stars at c. Nearly all blue stars down 
to sixth magnitude in the association I 
Ori have been taken in either of the two 
higher dispersions, most of them twice. 
The fainter stars measured at 4 A/mm 
are concentrated in a region of 1° diam- 
eter around the Trapezium. Analysis 
of this material is in progress. The gain 
in profile structure of the K line over that 
resolved by Adams is considerable ; on the 
average, the number of components seen 
is almost doubled, and as many as six 
components of K may be distinguished in 
e Orionis. With the highest dispersion, 
components separated by only 3 km/sec 
have been resolved. In some cases, very 
narrow components indicate a velocity 
dispersion of internal motions of not 
more than 1 km/sec. The correlation of 
the hydrogen and calcium profiles is a 
complicated problem. In the region of the 
Horsehead Nebula, hydrogen profiles are 
single-peaked and the K line shows four 
components. A similar situation applies 
in the area close to the Trapezium. 
Between rj and p Orionis, the 21-cm 
profile has several peaks. Further study 
will doubtless reveal a detailed picture of 
the cloud structure of the interstellar 
medium in the Orion region. 

In the region of the II Scorpii associa- 
tion, van Woerden obtained a plates for 
7, b plates for 13, and c plates for 7 stars. 
So far, hydrogen observations are avail- 
able in part of this area only. The 21-cm 
profiles appear to be more complicated 
here than those of the Ca II line. In 
Lacerta, 10 stars have been taken at b 

(2 also at a) and 15 at c. No detailed study 
of hydrogen in this association has yet 
been made, but observations at Dwinge- 
loo should be taken in the near future. 

In addition to stars in these three 
associations (and some foreground objects 
of early type), van Woerden has taken a 
plates of 23 and b plates of 16 bright 0-, 
B-, and A-type stars in other parts of the 
sky. These spectra will be used, in con- 
junction with Dwingeloo observations of 
neutral hydrogen at intermediate and 
high galactic latitudes, for a study of 
interstellar matter in the solar neighbor- 

The analysis of this extensive series of 
high-dispersion plates will include inter- 
stellar lines other than those of Ca II; 
and also, it is hoped, the strengths of 
stellar lines. 

Stimulated by the detection of the OH 
line of 18-cm wavelength at several radio 
observatories, Gaustad and van Woerden 
have attempted to determine the strength 
of the ultraviolet OH lines near X3080 in 
spectra of f Persei, P Cygni, and HD 
199579. These observations have not 
yet been fully assessed. 

Van Woerden has continued his 21-cm 
studies in collaboration with the staff 
of the Kapteyn Astronomical Laboratory 
at Groningen. Together with Dr. K. 
Takakubo of Sendai, Japan, he has 
further analyzed the line profiles of 
neutral hydrogen at intermediate galactic 
latitudes published in B.A.N., number 
524, 1962. A catalogue of gaussian com- 
ponents of these profiles has been pre- 
pared for publication. 

CN Lines 

The discovery by Munch of unusually 
strong interstellar absorption lines of 
CN in BD+66°1675, incorrectly re- 
ported last year as BD+60°2522, led to 
the observation of additional stars as- 
sociated with the dust-emission complex 
Sharpless 171. Two of them show the 
CN lines, and in one of them, BD+66° 
1674, the presence of lines arising from 
J = 2 and J = 3 is strongly suspected. 



On the basis of the evidence provided by 
the radial velocities and strengths of 
various interstellar lines in these and 
other foreground stars in the same 
region, an hypothesis has been advanced 
suggesting that the CN molecules pro- 
ducing the lines exist only in the H I 
region immediately surrounding the emis- 
sion nebula, and arise as a consequence of 
the exposure of dust in the H I region to 
an advancing ionization front. After a 
paper offering this suggestion had been 
sent for publication, it was found that 
the star HD 206267, imbedded in an 
emission nebular (IC 1396) obviously 
showing strong interaction with surround- 
ing dark clouds, also shows fairly strong 
CN lines, thus providing additional 
evidence in favor of the hypothesis. 

He I Nebular Absorption 

Munch has continued the observation 
of interstellar and nebular absorption 
lines in stars imbedded in emission 
nebulae. Observations of the stars in the 
clusters NGC 6523 and 6611, associated 
with the emission nebulae M 8 and M 16, 
have now been completed, bringing out 
again the fact that the strength of the 
nebular absorption of helium is not 
related to the apparent surface bright- 
ness. The measures of the intensity of the 
nebular emission lines and the contin- 
uum scattered by dust within the 
nebula, needed to test quantitatively the 
hypothesis that Lyman a may provide 
the predominant deactivation mechanism 
of the 2 3 jS He I level, by photoionization, 
have not kept pace with the spectro- 
graph^ work. However, it is planned to 
measure the reddening of the nebular 
light by photoelectric observations in the 
near future. 

Planetary Nebulae 

To gain further understanding of the 
observed optical-depth effects of the He I 
lines of the triplet states, the expansion 
velocities of the planetary nebulae spec- 
trophotometrically observed by O'Dell 
and not included in the 0. C. Wilson 

survey are being measured by Munch on 
multislit plates taken at Palomar. So far 
the nebulae IC 3568, NGC 4503, and 
NGC 6826 have been observed at 4.5 
A/mm dispersion. 

Interstellar Matter 

Gaustad, in collaboration with van 
Woerden, attempted to detect inter- 
stellar absorption lines of OH in the 
ultraviolet spectrum of f Per. No lines 
were detected, so that only an upper 
limit of 1.4 X 10 14 molecules per square 
centimeter could be set on the number in 
the line of sight. This number, combined 
with published measurement of the 21- 
cm hydrogen emission in the same direc- 
tion, implies an upper limit to the OH/H 
ratio of 2 X 10 -7 . The published value 
from the 18-cm radio observations is 
5 X 10~ 7 , but recent work suggests that 
the excitation temperature at radio 
frequencies is lower than assumed and 
that the radiofrequency spectrum value 
should be reduced. Gaustad has also been 
searching for other possible weak inter- 
stellar lines in the region longward of the 
ozone absorption bands; these lines, if 
real, have not been identified and have 
equivalent widths less than 6 mA. An 
attempt was made to identify ice in the 
interstellar grains. A search of the 
chemical literature on infrared absorption 
bands of solid H 2 0, CH 4 , and NH 3 per- 
mitted computations of the expected 
absorption bands in the reddened super- 
giant ix Cephei. They should have been 
very strong at 3.1 /z, owing to the funda- 
mental vibrational transition of water-ice 
in interstellar grains. Working with Drs. 
Danielson and N. Woolf of Princeton on 
the Stratoscope II balloon observations, 
Gaustad found that the 3.1-m band was 
not present. Consequently, not more than 
one-fourth of the interstellar reddening 
can be due to the ice particles. 

Whiteoak is using the spectrum scanner 
to obtain photoelectric observations of 
about 40 early-type stars to study the 
law of interstellar absorption in selected 
directions in the galactic plane. Groups of 



stars of spectral types 06-09 and with 
different amounts of interstellar absorp- 
tion are being observed in Perseus, Orion, 
Monoceros, Cygnus, and Cepheus. With 
a 50- A passband, measurements of the 
continuum of each star were obtained for 

30 regions between X3400 and X10800. 
The reduction of the observations is 
similar to that described above. For the 
blue spectral region, an RCA 1P21 photo- 
multiplier is used; for the red, an RCA 


Optical Observations of Quasi-Stellar 

The scanner has been used by Oke to 
study the continuous spectrum and emis- 
sion lines in the source 3C273. It is found 
that the emission lines of hydrogen and 
the optical continuous spectrum can be 
explained completely by a hydrogen 
nebula at an electron temperature of 
160,000°K. An alternative possibility is a 
hydrogen nebula at a temperature of 
about 15,000°K; a flat synchrotron spec- 
trum must be added, which accounts for 
more than 90 per cent of the optical 
continuous radiation. Hj3 is observed to 
be somewhat stronger than this theory 
predicts. In all models there is an excess of 
radiation in the infrared which is probably 
to be associated with the radiofrequency 
synchrotron radiation. The scanner was 
used to attempt to measure the polariza- 
tion of this infrared tail at XI 0000. 
Although the accuracy is still very poor, 
it is unlikely that the polarization at 
X10000 is as much as 10 per cent. Assum- 
ing that the high-temperature model is 
correct, we are led on the basis of the 
light variations to a nebula with a radius 
of, at most, a few parsecs and a mass of 
10 7 Mo. In any model of this size, 
electron scattering is important. At an 
electron temperature of 160,000°K, the 
emission-line widths may be caused 
entirely by electron scattering rather 
than by mass motions. Even if the 
line broadening is caused entirely by 
expansion, the total kinetic energy is 
sufficient for only 10 4 years. 

The discovery of the redshifts of 
3C273 and 3C48 mentioned in Year Book 

62 has led to many developments. The 
nature of these quasi-stellar radio sources 
has been investigated by Greenstein and 
Schmidt on the basis of all available 
radio and optical observations. Attempts 
to obtain additional redshifts, which are 
vital for a better understanding of these 
objects, are being made by Schmidt. 
Among the quasi-stellar sources investi- 
gated, redshifts were obtained for 3C47 
and 3C147, both of which were identified 
from radio positions obtained by Dr. 
T. A. Matthews. The redshifts, AX/X, are, 
respectively, 0.425 and 0.545. The source 
3C147 is thus the most distant object 
known in the universe, yet it is as bright 
as the 18th magnitude. Corrected for 
galactic absorption, its magnitude would 
be closer to 17. Schmidt has identified two 
well observed lines, X5760 [O II] and 
X5976 [Ne III], and three other less 
distinct lines, X4839 [O III], X5290 
[Ne V], and X6132 [Ne III]. In 3C47 
(redshift 0.425) he has identified seven 
observed lines: X3986 Mg II, X4885 
[NeV], X5310 [O II], X5510 [Ne III], X6200 
H T [O III], X7072 [O III], and X7136 
[O III]. 

The four quasi-stellar radio sources 
with known redshifts (3C47, 48, 147, and 
273) show, both in radio luminosity and 
in optical luminosity, a range of a 
factor of about 10. Their very bright 
absolute magnitudes (about —25 visual) 
make these objects observable at larger 
distances than ordinary galaxies or radio 
galaxies. There is a wide range of both 
radio properties (linear size, structure, 
luminosity, and spectral index) and 
optical properties (emission lines observed 
and absolute magnitude) among the four 



quasi-stellar sources having known dis- 
tances. The radio properties are at best 
only slightly different from those of the 
more normal radio galaxies; the optical 
properties are entirely different. 

Unfortunately, the contrast of emission 
lines against the continuum in quasi- 
stellar objects is rather poor. Lack of in- 
formation about the far ultraviolet emis- 
sion spectrum to be expected in these 
sources is another reason why the de- 
termination of larger redshifts is not as 
easy as the great optical luminosity of 
these objects would perhaps suggest. 

The program of optical identification 
of radio sources of the quasi-stellar type 
was continued by Sandage. The discovery 
technique is based on the intense ultra- 
violet radiation of these objects as com- 
pared with the blue and yellow emission 
in the F(X) d\ distribution law. The 
observing program consists of photo- 
graphing each radio source position of the 
revised 3C catalogue with the 100-inch 
reflector or the 48-inch schmidt. Double- 
image photography utilizing a blue and 
an ultraviolet filter for successive ex- 
posures allows the ultraviolet objects to 
be found by inspection. A tentative 
identification can be made when the 
ultraviolet star image occurs within the 
error rectangle of the radio position. 
Photoelectric observations are then made 
with the 200-inch, and a unique identi- 
fication is possible based on the known 
photometric properties of the quasi- 
stellar objects. 

New identifications using this tech- 
nique during the report year include 3C9, 
3C93, 3C208, 3C216, 3C228, 3C245, 
3C287, and possibly 3C263.1. These to- 
gether with 3C47, 3C48, 3C147, 3C196, 
3C273, and 3C286 found during the past 
several years by various members of the 
staff provide the list of definitely identi- 
fied quasi-stellar radio sources available 
at the end of the report year. 

All the sources except 3C228 and 
3C263.1 have been observed photo- 
electrically at least once, and all objects 
exhibit peculiar B— V to U—B ratios in 

comparison with normal stars. The data 
all fall close to the optical synchrotron 
line in the B—V, U — B diagram, but no 
clear evidence yet exists to show that the 
optical radiation is indeed caused by the 
synchrotron process. A blackbody distri- 
bution or optically thin bremsstrahlung 
(free-free transitions) can also fit part of 
the data. 

Identification with Galaxies 

During the course of the observational 
program, Sandage identified 3C305 with 
an Sa galaxy and 3C275.1 with a peculiar 
Sc galaxy. Spectra of 3C305 taken with 
the 200-inch prime-focus spectrograph 
show a bright X3727 emission line which 
in certain position angles of the slit is not 
coincident with the nucleus. A series of 
spectrograms in different position angles 
shows that the X3727 emission has a 
peculiar distribution across the face of 
the galaxy, suggesting a jet of emitting 
gas coming from the nucleus in a type of 
spiral-arm pattern. The 3C275.1 radio 
position is coincident with a strong jet 
emerging from the spiral galaxy NGC 
4651. Limiting direct exposures with the 
48-inch schmidt show that the jet is 
accompanied by a counter jet on the 
other side of the galaxy. Both these 
filaments terminate on the boundary of an 
extended halo which surrounds the entire 
optical image. The jets are not polarized 
in optical radiation to within 10 per cent. 

The entire 3C catalogue will be sur- 
veyed in the current program, and it is 
hoped that large-scale reflector plates of 
the interesting objects will be obtained 
so that a search may be made for peculiar 
features. As part of this program, special 
polarization plates of the optical jet from 
3C273 were taken by Sandage and by Arp 
with the 200-inch; these data show that 
no linear polarization greater than about 
10 per cent exists in the jet. 


The exciting problem of the physical 
nature of the quasi-stellar radio sources 



has been extensively discussed. The in- 
terpretation of the observations of these 
objects is extremely puzzling and will not 
be clarified without very much further 
observational study and theoretical 
analysis. In the model that can be most 
directly derived from observation, accord- 
ing to Greenstein and Schmidt, the radio 
disturbances come from a relatively 
large region surrounding a small gaseous 
nebula, of the order of 1 parsec radius for 
3C273, possibly somewhat larger for 
3C48. The continuous spectrum in 3C273 
may be a mixture of hydrogen recom- 
bination continua, two-photon emission, 
and synchrotron radiation. In 3C48 it 
seems to be largely synchrotron radiation. 
The emission lines in 3C273 are largely 
hydrogen; in 3C48, which seems to be a 
less dense object of higher excitation, 
largely forbidden lines. However, the 
strongest line in 3C48 is the resonance line 
of ionized magnesium. From the weakness 
of the [0 II] line in 3C273, an electron 
density near 5 X 10 6 cm -3 is derived. The 
light variations suggest that the size is 
small, which is consistent with high 

The density is so large that radio waves 
cannot be produced within this ionized 
region of high density, and, if synchro- 
tron emission is produced, the high- 
energy electrons are very quickly de- 
stroyed by the inverse Compton effect. 
Various models for the synchrotron 
radiation suggest that, in fact, the high- 
energy electrons may cause the heating 
and ionization of the gas that we observe. 
The detailed spectroscopic analysis of the 
emission-line spectrum permits mass de- 
terminations (about 10 6 solar masses) 
and also determination of the total 
energies available. 

Such objects as the jet in 3C273, 
analyzed in the conventional manner for 
minimal energy as a function of magnetic 
field, have enough energy for long periods 
of time, even at their enormous luminos- 
ity, but the same is not true for the small 
ionized regions. Their internal energy, if 
it is only the ionization, 13 volts per 

electron-proton pair, is sufficient to 
maintain their luminosity for about 10 
days. If the kinetic energy is correctly in- 
dicated by the line width, and is the energy 
source, the lifetime is only some 30 years. 
Consequently, it is necessary to replenish 
the energy of these ionized regions, pre- 
sumably from an internal source. It seems 
most likely that some invisible smaller 
object is the source of energy output. If 
its gravitational mass causes the line 
broadening observed, a mass of at least 
10 8 suns is required in an invisible object. 
Several interesting astrophysical problems 
are brought up by these H II regions. 
The optical thickness is so great, reaching 
10 7 in the ionized magnesium resonance 
doublet, that a special transfer theory 
needs to be derived. The auroral lines 
of [Ne V], present in 3C48, determine 
the minimum electron temperature as 
about 15,000°. The optical depth for 
electron scattering is appreciable, so 
that the quanta are diffused by free 
electrons, further slowing down any 
intrinsic light variations that may exist. 
The light variation, with its required 
very small source size, confirmed in this 
investigation by the high electron 
density derived, remains an outstanding 

Greenstein has studied the properties 
of synchrotron radiation at optical ener- 
gies if the critical frequency for synchro- 
tron radiation is in the optical ordinary 
ultraviolet region. It seems difficult to 
obtain sufficient ultraviolet ionizing 
energy from even the very high-energy 
electrons required for optical synchrotron 
emission. Only if there is a high-energy 
tail of especially energetic electrons can 
the ionization levels near 100 electron 
volts be obtained. The bolometric correc- 
tion for synchrotron radiation is a func- 
tion of the critical frequency; for a 
critical frequency near 10 14 sec -1 , the 
bolometric correction is about 5 magni- 
tudes, dropping to a minimum of about 1 
magnitude at a critical frequency of 10 15 
sec -1 , and rising again above 2 magni- 
tudes at 10 16 sec -1 . 




Magnitudes and Redshifts 

Baum has continued his multicolor 
photoelectric observations of galaxies 
belonging to distant clusters. These 
measurements yield both the redshifts 
and the relative bolometric magnitudes in 
the manner described from time to time 
in earlier annual reports. The redshift- 
magnitude relation has thus far been the 
best available test for distinguishing 
among various cosmological models. Ob- 
servations during the report year provide 
magnitude data for thirty galaxies be- 
longing to three clusters of known red- 

Schmidt continued a program aimed at 
obtaining optical spectra and redshifts of 
galaxies identified with radio sources. 
Three of the newly observed radio 
galaxies (3C435, 456, and 459) have red- 
shifts between 0.21 and 0.24. A paper is 
being prepared giving redshifts and a 
qualitative description of the spectra of 
sufficiently observed radio galaxies. There 
is suggestive evidence that radio galaxies 
with strong emission lines show less or no 
absorption in the optical continuum; 
observations of the continuum, however, 
are very difficult in these faint galaxies. 

Investigations of Individual Galaxies 

To obtain a reference system for the 
interpretation of the radial velocities 
measured from the emission lines of [0 II] 
present in the central regions of M 31, 
Munch has redetermined the rotational 
velocity of the diffuse stellar light from 
the same regions. Spectrograms at 196 
A/mm dispersion, with slits in the direc- 
tion of the minor axis of the system, and 
at distances at either side from the 
nucleus of 10', 8', 5', and 3' on the major 
axis, have been obtained with the 
Palomar nebular spectrograph. The differ- 
ences in radial velocity between the 
various points on the major axis and the 
nucleus appear to increase monotonically 

with distance to the nucleus, and there is 
no indication of a secondary minimum 
at a distance of about 9', present in the 
earlier measures of H. W. Babcock. 

Before his retirement, M. L. Humason 
obtained the radial velocities of a number 
of emission regions in the Sc galaxy NGC 
2403 (Year Book 52). Because the individ- 
ual radial velocities then found appeared 
to deviate from a mean rotation curve by 
amounts considerably greater than what 
would be expected from the accuracy of 
the measures, however, these results were 
not published at the time. During the last 
two years, ten of the brightest emission 
regions have been reobserved by Munch, 
mainly in Ha and [N II], at a dispersion 
of 110 A/mm. The agreement with 
Humason's results has been found satis- 
factory, and it has been inferred that the 
dispersion of the velocities around a 
mean curve is real, probably a result of 
the lack of coplanarity of the various 
emission regions. 

Many of the galaxies in the Virgo 
cluster have sufficiently large redshifts 
to prevent the intrinsic D lines in their 
spectra from being contaminated by 
airglow emission at D. For eleven of these 
galaxies that are ellipticals, Deutsch has 
now obtained spectrograms at 230 A/mm. 
These spectrograms show the galaxian D 
lines to range downward in strength from 
about 10 A equivalent width in NGC 
4649 to about 6A in NGC 4958. The 
actual width of the blended doublet 
varies independently of the equivalent 
width, being greatest in NGC 4486. In 
that object, a well known radio galaxy 
which exhibits optical synchrotron radia- 
tion, the D lines may be filled in by the 
synchrotron continuum. Deutsch also 
finds that a source of continuous radiation 
lies in the "stellar" nucleus of the Seyfert 
galaxy NGC 1068, which appreciably 
weakens the absorption lines at D as 
compared with their strength several 
seconds of arc away from the nucleus. 
Deutsch plans to continue his observa- 



tions of the strength and width of the D 
lines in galaxies of various types and 

Miss Swope has continued work on the 
fields 15' and 50' of arc south-preceding 
the nucleus of the Andromeda galaxy 
M 31. This involves a discussion of 200 
cepheids and 15 variables of population 
II. The cepheids seem to be like those of 
our Galaxy, and there are differences 
between them and those of the Magellanic 
Clouds. The long-period cepheids tend to 
appear closer to the nucleus, and shorter- 
period cepheids away from the nucleus. 

Since the end of January 1964, Miss 
Swope has been on leave at the Royal 
Greenwich Observatory, where she has 
been preparing data for a color-magni- 
tude diagram of stars in the spiral arms 
of M 33. The measures are made on 200- 
inch telescope plates taken by Sandage. 
This work is about half completed. 
Reduction of the measures will require 
adding more faint stars to the photo- 
electric sequence. 

It was reported last year that M 82 is 
a galaxy undergoing a violent explosion 
of unknown origin in its central region. 
New data obtained during the report 
year by Sandage and Miller now provide 
rather direct evidence for the presence of 
a large-scale organized magnetic field in 
M 82 which extends along the minor axis 
of the galaxy. An extensive outer system 
of blue filaments exists in this galaxy, 
reaching to distances of 4000 parsecs 
above and below the fundamental plane. 
The light from these filaments is con- 
tinuum radiation and is found to be 
linearly polarized with the electric vector 
parallel to the major axis of M 82. 
The percentage of polarization is very 
high, reaching perhaps 80 in parts of the 
filaments. The observations can be ex- 
plained if the light from the filaments 
originates from synchrotron emission of 
relativistic electrons or positrons or both 
moving in an organized magnetic field. 
Such emission is expected from theory to 
be linearly polarized, with the electric 
vector perpendicular to the magnetic 

field and to the motion of the charged 

The energetics of M 82, calculated from 
the 1963 model of M 82 by Sandage and 
Lynds, suggest that the average magnetic 
field lies between 10~ 5 and 10 -6 gauss. 
These limits on the field strength require 
that the electrons have energies between 
2.5 X 10 12 and 8 X 10 12 electron volts if 
they are to radiate synchrotron energy at 
optical frequencies (i.e., v > 5 X 10 14 
cycles per second). Such high energies are 
in the cosmic-ray range and provide 
further evidence that the origin of cosmic 
rays throughout the universe is connected 
with the violent events occurring in radio 
galaxies. Three-color photoelectric sur- 
face-brightness measurements of patches 
of the blue filaments in M 82 give colors 
ranging from B— 7 = 0.47, U — B = 
-0.40 to B- V = 0.65, U-B = -0.36. 
These colors agree satisfactorily with 
calculated colors obtained from the theo- 
retical synchrotron spectrum of the form 

F(v) = 4.3 X 10- 

y-0.23 / 


vlv ( 

This spectrum fits the radio data, and it 
extends correctly into the optical region, 
as judged from the measured colors, if the 
critical frequency v c (which corresponds 
with the high energy cutoff of the elec- 
trons) is 5 X 10 14 cycles per second. 
Whether the theoretical spectrum with 
such a low cutoff frequency can provide 
enough ionization energy below the 
Lyman limit to account for the intensity 
of the hydrogen recombination spectrum 
in the Ha expanding filaments must be 
tested next year by absolute photom- 
etry of the total emitted flux in the 
hydrogen lines. 

Photometry of the variable stars in the 
Sc galaxy NGC 2403 was continued by 
Tammann using the long series of 200- 
inch plates obtained mostly by Sandage 
since 1949. Secondary photometric stand- 
ards were set up in V and B magnitudes 
based on 120 stars spread over the face 
of the galaxy for which old and new 



photoelectric measures by Sandage are 
available. Sixteen cepheids have now been 
found with periods ranging from 20.23 to 
87.48 days. The period-luminosity rela- 
tion can be obtained only for 5 max be- 
cause of the faintness of the variables 
(£ m ax lies between 21 m 25 and 22 m 81). 
The slope of the period-luminosity func- 
tion cannot be determined from so few 
stars, but preliminary comparisons with 
galactic and extragalactic cepheids of 
known distance seem to confirm a dis- 
tance modulus of about 27?50. 

Four irregular variables of intermediate 
color reach J5 max values brighter than 
20^00 . The brightest of these stayed near 
magnitude 21 from 1910 to 1949, but 
brightened in 1949, reaching B = 16.3 in 
November 1954. This magnitude corre- 
sponds to an absolute magnitude of Mb = 
— 11.2 uncorrected for absorption, and, 
therefore, the star is one of the brightest 
known in external galaxies. The star has 
now faded below B = 22 m . Fourteen ad- 
ditional variables of intermediate color 
are known, some of which may well be 
cepheids whose periods cannot yet be 
found. Thirteen more stars are known to 
be irregular red variables of small ampli- 
tude. Two possible novae are known, but 
since each is visible on only one plate 
their reality is uncertain. 

A photoelectric sequence has been 
measured in M 101 to B ~ 22 m by San- 
dage in order to determine the light curve 
of the explosive star discovered in 1951 
by Hubble. Tammann has derived the 
light curve starting with the discovery 
plate of February 3, 1951, and continuing 
to the last plate on which the star is 
visible in June 1952. The brightness faded 
from B = 17 m 5 (M B c~ — lO^O uncor- 
rected for absorption) on the first plate, 
becoming fainter than B = 22 m on the last 
one. The average gradient of the light 
curve corresponds to only 0.009 magnitude 
per day, and the decrease in brightness by 
3 magnitudes took more than 300 days. 
This slow rate of decline excludes the 
possibility of a very bright normal 
galactic-type nova, whereas the maximum 

brightness seems too faint by at least 4 
magnitudes for a supernova even of type 
II, a type that would otherwise be sug- 
gested by the position in the middle of 
a spiral arm and by the long decay time. 
Although it is true that the actual maxi- 
mum brightness was not observed be- 
cause it occurred before M 101 could be 
observed in the 1950-1951 season, the 
light curve gives no indication that the 
star could have been 4 or more magni- 
tudes brighter, as would be required for a 
supernova of type II. The star, therefore, 
seems to be an explosive object of a 
heretofore unknown type. 

Blue-sensitive photographs of a sample 
of dwarf galaxies, mostly selected from 
van den Bergh's catalogue of 222 such 
objects derived from Sky Survey plates, 
have been obtained. Rough estimates of 
the apparent magnitudes of the brightest 
stars in 12 of these dwarf galaxies indi- 
cate that within 6 million parsecs the 
average number density of dwarfs of this 
kind is of the order of 10 -19 per cubic 
parsec. This corresponds to an average 
mass density of about 3 X 10~ 33 g/cm 3 . 

Work by Arp and Dr. A. D. Thackeray 
continues on the young spherical cluster 
NGC 1866 in the Large Magellanic Cloud. 
Most recently, B and V light curves were 
obtained for eight cepheids within the 

All photographic observations by 
Hubble, Baade, Sandage, Arp, and others 
of the dwarf galaxy NGC 6822, a member 
of the local group, and a photoelectric 
sequence (by Arp) have been turned over 
to Susan Kayser, who is analyzing the 
material for her doctoral dissertation. 

Peculiar and Interacting Galaxies 

More than 250 limiting photographs 
have been obtained by Arp with the 200- 
inch telescope in compiling an atlas of 
peculiar galaxies. Unusual galaxies from 
published lists by Vorontsov-Velyaminov, 
W. W. Morgan, and Zwicky and from 
unpublished lists by A. G. Wilson, 
Wirtanen, Page, Herzog, and others were 
examined on 48-inch Sky Survey prints, 



only the most interesting objects being 
selected as candidates for the atlas. Some 
entirely new objects were found in the 
processes of checking the preliminary list 
on the Sky Survey prints. Photographic 
copying of the plates is scheduled for the 
coming summer. 

Classification of the objects is in a 
preliminary stage at present, but some of 
the most interesting classes that are 
apparent are: (1) galaxies appearing to 
show ejection of material; (2) spirals 
with a companion at end of an arm (M 51 
class); (3) spiral arms that are multiple, 
branched, bent, twisted, or ruptured; 
(4) galaxies with long streamers or fila- 
ments sometimes connecting pairs, some- 
times not; (5) E galaxies with diffuse 
plumes; (6) pairs with cometlike tails; 
and (7) small condensed knots within 
galaxies which show tails or trails. 

Faint Features in Galaxies 

With a plate and filter combination 
that isolates the darkest spectral region, 
the dark skies of sunspot minimum are 
being used by Arp to photograph the 
faintest observable luminous features in 
galaxies. Three or more plates with the 
48-inch schmidt are printed together to 
improve the image-to-grain ratio for very 
faint features. William Miller has been 
developing these printing techniques. 
The //2.5 schmidt requires at least 3 
hours' total exposure on such objects 
(corresponding to prohibitively long ex- 
posure times with the 200-inch), but with 
this plate and filter combination it goes 
more than a magnitude fainter than 
normal sky-limited photographs. Features 
that have never been seen before are 
emerging in the faint outer regions of 
galaxies. So far some of the most interest- 
ing results are : 

1 . Fainter outer extensions of the radio 
galaxy NGC 1316 (Fornax A) curve over 
into the hitherto "empty" regions of 
radio emission. Since radio polarization is 
roughly perpendicular to these extensions, 
it is suggested that a magnetic field runs 
along them analogous to a spiral arm. 

The important question of the source of 
the radiation being observed remains to 
be solved. 

2. A faint luminous filament is seen to 
encircle the end of M 81 lying closest to 
M 82. There is a suggestion of faint 
luminous material lying between M 81 
and NGC 3077, the companion on the 
other side. 

3. Faint extensions on M 31 and NGC 
205 appear on photographs of the M 31 

The Leo group, certain dense clusters 
of galaxies, and other objects are also 
under study with this technique. 

Interference-Filter Photography 

Arp is using the 48-inch schmidt with 
[O II] and Ha interference filters to study 
the distribution of the sources of these 
emissions throughout galaxies, particu- 
larly in spiral galaxies. What appears to 
be an eruption or blowout in one of the 
M 51 spiral arms is outlined clearly in the 
light of [0 II]. Faint extensions on NGC 
5195, the companion to M 51, have been 
discovered. One feature extends more 
than a diameter of the object itself and 
shows up best at a wavelength just short 
of Ha. 


The 200-inch telescope with a polariz- 
ing filter was used by Arp to photograph 
3C273, once with the axis aligned along 
the jet and again with it perpendicular to 
the jet. The overall polarization was no 
more than 10 per cent; therefore, if 
appreciable optical synchrotron radiation 
is present in the jet, the magnetic field 
cannot be aligned over dimensions com- 
parable to those of the jet. There are 
indefinite indications that smaller features 
within the jet may show polarization, 
implying a structure similar to that of 
the jet in M 87. 

Occasional polarization photographs 
of galaxies are being made with the 48- 
inch in attempts to detect objects showing 
a high degree of polarization. Only the 
galactic nebula NGC 1999 in Orion, 



however, has shown "strong" polariza- 
tion (E vector tangent to the rim). 

Compact Galaxies 

Zwicky has adduced reasons for the 
existence of galaxies more compact than 
any investigated so far, as well as massive 
compact parts of galaxies. During the 
past few years he has made a concerted 
effort to check on these predictions 
through the investigation of objects 
which on 48-inch schmidt plates often 
appear like stars but which, on closer 
inspection, are suspected of being extra- 
galactic stellar systems. Sometimes, how- 
ever, when photographed with the 200- 
inch telescope or when investigated 
spectroscopically, such objects prove to be 
double stars or individual stars projected 
on very faint distant galaxies. 

So far the search has been limited to 
moderately compact and to compact 
galaxies. On the average, about 1 such 
galaxy has been found per 5 square 
degrees on 48-inch schmidt plates, of 
apparent luminosity brighter than that 
corresponding to the photographic mag- 
nitude m p = +17.0. Several hundred of 
these objects have been selected for 
further detailed study. About thirty 
compact galaxies have already been 
photographed directly and investigated 
spectrographically with the 200-inch tele- 
scope. Some of the results obtained are as 
follows : 

1. Although most compact galaxies 
appear as isolated objects, many of them 
occur in groups of two, three, or more. 
Some show exceedingly faint extensions, 
such as jets, halos, and spiral arms. 
Particularly interesting are compact 
galaxies that contain several nuclei 
imbedded in a luminous amorphous 
substratum. Their importance lies in the 
fact that both the width of the spectral 
lines, within the nuclei and the sub- 
stratum, and the relative displacements 
of these lines, allow independent esti- 
mates of the masses to be made, provided 
that sufficiently strong observational 

evidence exists that these galaxies are 
statistically stationary systems. 

2. The spectra of the compact galaxies 
so far investigated range in type from 
K to 0, some without and others with 
emission lines superposed on continua of 
various colors. A few show emission lines 
only. Either allowed or forbidden lines, or 
both, may be present. The percentage of 
early-type spectra among the compact 
galaxies is high in comparison with 
ordinary galaxies. 

3. The Doppler widths of both the 
absorption and the emission lines may 
vary from a few to many thousands of 
kilometers per second, the larger widths 
indicating either high internal motions 
and, therefore, high mass and compact- 
ness, or possibly, for exploding systems, 
very rapid expansion. 

4. The absolute luminosities of the 
systems investigated so far range from 
10 8 to 10 10 times the luminosity of the 
sun, and the masses are in the range from 
10 10 to 10 12 suns. The indicated mass- 
luminosity ratios lie between 50 and 500 
— very much higher than those of 
ordinary galaxies. 

5. The symbolic velocities of recession 
of the compact galaxies so far studied lie 
in the range from 1000 to 30,000 km/sec. 

6. Pronounced compact nuclei and 
other very compact regions are not neces- 
sarily found in the brightest and largest 
galaxies. For instance, the brightest 
galaxies in the Coma cluster do not 
contain any such nuclei. To evaluate 
statistically the occurrence of nuclei and 
compact stellar aggregates in cluster 
galaxies, a series of photographs with in- 
creasing exposure time was taken with 
the 48-inch and 200-inch telescopes. It 
was found that compact regions of very 
high surface brightness exist with rela- 
tively high frequency within galaxies of 
medium total luminosity. 

7. The gravitational potential of some 
of the compact galaxies is so high that 
light passing along their surfaces may be 
expected to be deflected by as much as 1 
minute of arc, if the general theory of 



relativity is correct. A strenuous effort is 
therefore being made to discover both 
double and ringlike images of remote 
galaxies which are generated by the 
gravitational lens effect due to a compact 
galaxy lying nearer and in the line of 
sight with the remote galaxy. 

8. Unusual spectra are found for some 
compact galaxies that show jets and other 
protuberances. In these spectra, emission 
lines are always conspicuous and their 
line widths indicate expansion velocities 
of several thousand kilometers per second 
— one of more than 10,000 km/sec. 
Typical examples are spectra of the 
anonymous galaxies at a = ll h 22 m 7, 
8 = 54°10',and a = l h 19 m 5, 8 = -1°18' 
(1950). In the former the widths of the 
emission lines of hydrogen are about 4 
times those of the (forbidden) emission 
lines of oxygen. 

Catalogue of Galaxies and of Clusters 
of Galaxies 

Volume II of the Catalogue was 
published in the fall of 1963. It contains 
data on about 6000 galaxies brighter 
than the apparent photographic magni- 
tude + 15.7 and on about 2500 rich 
clusters of galaxies. It covers 89 of the 
48-inch schmidt Sky Survey fields in the 
zones +18°, +24°, +30° in declination 
and in the range of right ascension from 
6 h 30 m to 18 h 30 m . Volume III, which will 
cover 111 of the fields in the zones +36°, 
+42°, +48°, and +54° in declination and 
intervals in right ascension from 5 h to 
20 h , is in preparation. All the work on 
clusters of galaxies for this volume has 
been completed. Volume V, covering 102 
fields in the declination zone 0°, +6°, 
+ 12°, and +18° in the range of right 
ascension from 20 h to 6 h , is also in prep- 
aration. The positions and apparent 
photographic magnitudes of all the indi- 
vidual galaxies to be incorporated in this 
volume have already been measured. 
Some preparatory work has been done on 
the compilation of volumes IV and VI. 

Clusters of Galaxies 

Zwicky and Mrs. J. Berger have com- 
pleted the analysis of the sky coverage 
by clusters of galaxies, using the data 
contained in volume II of the Catalogue. 
All essential results check those obtained 
by Zwicky and Rudnicki in their anal- 
ogous analysis of the data listed in 
volume I. This analysis will be repeated 
separately for each of the six volumes as 
they are completed. 

To arrive at a definitive answer to the 
problem of the apparent and the absolute 
size of the large and rich spherical clusters 
of galaxies, Zwicky is engaged in follow- 
ing the equal population contours (iso- 
pleths) to the lowest depressions sur- 
rounding the large clusters. An extended 
depression of this type northeast and 
northwest of the Coma Cluster has now 
been located as a result of the work 
done on the first three volumes of the 
Catalogue. By determining the isopleths 
from the depression outward and ex- 
trapolating to the isopleths that were 
determined by proceeding outward from 
the center of the Coma Cluster, it is 
hoped that a final apparent diameter for 
this cluster can be found. 


In the period from May 31, 1963, to 
May 31, 1964, twelve supernovae were 
discovered at Palomar (one by H. S. 
Gates, five by C. Kowal, three by G. 
Reaves, and three by Zwicky). In the 
same period, eight supernovae were dis- 
covered by other members of the Com- 
mittee for Research on Supernovae 
(Commission 38, I. A. U.) working at 
observatories in Europe and Mexico. 
Whereas most of the supernovae found 
at Palomar are located in galaxies of 
distant clusters, most of the eight dis- 
covered at European and Mexican observ- 
atories are in relatively close nearby 
galaxies and are particularly valuable for 
detailed studies. Spectra of about half of 



all 19 supernovae as well as spectra of 
their parent galaxies have been secured by 
Zwicky with the nebular spectrograph of 
the 200-inch telescope. Light curves have 
been determined by Zwicky and by 
Zwicky and M. Karpowicz for the super- 

novae in NGC 1058 (type V), NGC 2841 
(type I), and NGC 3992 (type I). 

A monograph intended as a compila- 
tion of all essential data on the 151 
supernovae discovered since 1885 is in 


Photography of fields in the declination 
zones —36° and —42° has been under- 
taken by Whiteoak with the 48-inch 
Palomar schmidt telescope. The region 
of sky covered is farther south than the 
southern extension of the National Geo- 
graphic Society -Palomar Observatory Sky 
Survey. Exposures are 20 minutes, with 
103a-E emulsion and amber Plexiglass 
filter. With this combination, transmis- 
sion has been increased 50 per cent over 
that of the red combination of the 
Survey. Maximum response occurs at 

X6500; the response is half maximum at 
X6000. This survey contains 100 centers. 
The plates are not of high quality, since 
at large zenith distance the images suffer 
from the degrading effects of frequent 
poor seeing and of the differential refrac- 
tion across each plate during exposure. 
They are intended only for temporary use 
until such time as they can be superseded 
by a sky survey carried out by an ob- 
servatory in the southern hemisphere. A 
limited number of sets of prints will be 
made available to southern observatories. 


Arne Wyller completed a study on the 
thermal conductivity tensors for a non- 
relativist ic, degenerate hydrogen gas in 
the presence of a magnetic field. By in- 
cluding the proper velocity dependence 
of the electron-ion collision frequency, he 
found orders-of-magnitude differences for 
the conductivity as compared with values 
derived on the assumption of a constant, 
velocity-independent collision frequency. 
Thus the magnetic field has great effi- 
ciency in affecting thermal conduction 
of instability in degenerate stellar cores, 
since the energy could be transported 
by conduction along the field and by 
radiation perpendicular to it. Another 
cause of instability could reside in the 
regions of partial degeneracy, where the 
Hall conductivity apparently reverses 
sign as the degeneracy increases. 

Arpigny initiated a project for the 
construction of stellar model atmospheres, 
using the California Institute of Tech- 

nology stellar-atmosphere computer pro- 
gram, suitable for the study of the atmos- 
pheres of hot subdwarfs and of white 
dwarfs. The range of log g is 5 to 8, in 
effective temperature, from 10,000° to 
35,000°K, with different hydrogen-to- 
helium ratios. 

The effects of the continuous absorp- 
tion of the HeH + quasi-molecule on the 
structure of model atmospheres of early 
type has been studied by R. H. Norton 
as a Ph.D. thesis project. On the basis of 
approximate radiative cross sections cal- 
culated from s orbitals of the Slater type 
and classical statistics for the inter- 
nuclear separations, Norton finds that 
the maximum effect of the HeH + absorp- 
tion takes place for an effective tempera- 
ture of 16,000°K. For a main-sequence 
atmosphere of this temperature, the 
contribution of HeH+ to the opacity is 
comparable to that provided by hydrogen 
absorption in the Balmer continuum at 



wavelengths below threshold, near 1130 A. 
The absorption of HeH + , therefore, does 
not significantly affect the radiative 
equilibrium of stellar atmospheres. 

With Mr. James Bardeen, Deutsch has 
studied a simple theoretical model for the 
formation and gravitational collapse of a 
supermassive object near the center of a 
spherical galaxy. In this model it is 
assumed that each star in a massive 
spherical galaxy continuously injects gas 
into the interstellar medium at about the 
rate estimated for mass loss by the sun in 
the phenomenon of the quiet-sun solar 
wind. Unless the interstellar gas is ex- 
tremely hot, or endowed with appreciable 
net angular momentum or magnetic 
energy, it cannot be supported against the 
gravitational field arising from the stars in 
the galaxy. Calculations have been made 
to describe the spherically symmetric 
collapse of the interstellar gas that must 
then ensue. These calculations indicate 
the possibility that matter lost from 
stars could form, near the center of a 
galaxy, a supermassive condensation of 
10 6 -10 8 solar masses, on a time scale of 
about 10 8 years. Such a condensation 
may be associated with a quasi-stellar 
object, and in its later stages it may pro- 
duce the kind of relativistic implosion 
that has been described by Fowler, 
Hoyle, and others. In the models of 
Deutsch and Bardeen, the gas flows have 
been severely idealized, and much re- 

mains to be done if they are to be useful 
in elucidating the phenomena observed 
in the quasi-stellar sources. 

A theoretical investigation has been 
initiated by Gunn on a very general 
semiempirical approach to the statistical 
structure of the distribution of galaxies; 
it attempts to avoid both the simplifying 
assumption of continuity of the distribu- 
tion of matter and ad hoc "clustering" 
models. Use is made of a description due 
to Layzer in the course of error analysis, 
but it need not be assumed that the dis- 
tribution is strictly of the type described 
by him. The analysis is of the "spectral 
theoretical" type, which has been very 
successful in recent years in communica- 
tion theory, and bears superficial resem- 
blance to the method of autocorrelation 
analysis. The theory is being applied both 
to counts of nebulae and to the analysis 
of the fluctuations in the background 
cosmic light due to nebulae too faint to be 
seen as such. Cosmological parameters 
enter in a straightforward way, and it is 
hoped that the theory will provide a new 
cosmological test. Mathematical formula- 
tion of the theory is essentially complete, 
and a small pilot study (on the distribu- 
tion of counts) is under way to provide 
the necessary data for an efficient full- 
scale investigation; plates for the pilot 
study have been obtained with the 
48-inch schmidt camera. 


An electronics instrumentation group 
has been formed, under the supervision 
of E. W. Dennison, consisting of John 
Shirley, engineer, and L. Blake£, M. 
Scheutz, and B. Smith, technicians; it 
provides the Observatories' staff members 
with organized capability for the design, 
construction, modification, and mainte- 
nance of electronic instruments. Rapid 
design and construction of new "state of 
the art" instruments of unusually high 
reliability is the goal of this group, which 
is supported jointly by the National 

Aeronautics and Space Administration, 
the California Institute of Technology, 
and the Carnegie Institution of Wash- 

Data Acquisition 

A new universal pulse-counting and 
data - recording system for acquiring 
photoelectric data at the 200-inch tele- 
scope has been designed and assembled. 
Though designed initially for use with 
the prime-focus scanner previously built 
for J. B. Oke (see below), the system can 



be operated with other photometric appa- 
ratus. It consists of a prime-focus cage 
unit for the Hale telescope, a reversing 
counter and data-control unit, and a 
digital data recorder. 

The prime-focus cage unit is assembled 
almost entirely from standard compo- 
nents manufactured by the Radiation 
Instrument Development Company, in- 
cluding a preamplifier, a double delay- 
line amplifier, a discriminator, a rate 
meter, and a high-voltage power supply. 
This instrument shows the expected in- 
sensitivity to amplifier gain and high- 
voltage changes. Moreover, locating the 
discriminator in the prime-focus cage per- 
mits 10-volt pulses to be transmitted to 
the data room, thereby essentially elimi- 
nating the sensitivity to external electri- 
cal noise sources. Low multiplier-tube 
gains are also possible, permitting high 
count rates at low average anode cur- 
rents. In one example, a tube gain of 
1.65 X 10 5 was used. This means that an 
average count of 100,000 counts per sec- 
ond corresponds to 2.7 X 10 -9 ampere 
anode current. A coincidence correction 
of approximately 10 per cent was ob- 
served at this counting rate. In addition 
to operation as a photon-counting pho- 
tometer, the counter may be operated in a 
reversing or up /down mode. A synchro- 
nizing signal generated by a photodiode, 
exposed to a light through the chopper 
wheel on the prime-focus scanner, con- 
trols the direction of count. The photon 
count from the night sky can be sub- 
tracted from the star and sky count, 
leaving only the photon count due to the 
star alone. This instrument combines the 
stability and sensitivity of phase detec- 
tion with the advantages of photon 

The data are recorded on a printed paper 
tape as well as a printing summary punch. 
Numerical indicator lamps provide in- 
stantaneous visual data presentation. The 
reversing counter and data-control cir- 
cuits were all constructed from transistor 
data logic circuit elements purchased 
from the Decisional Control Company. 
Several of these elements are assembled 

on each removable card or board. This 
type of construction facilitates mainte- 
nance and circuit modifications. A more 
comprehensive data-collection system has 
been designed as an adjunct to the equip- 
ment described above. 

150-Foot Solar Tower 

A second major project undertaken 
this year by Dennison's group was the 
design of a solar guider, raster scan, and 
data-acquisition system for the 150-foot 
solar tower. The design requirements in- 
clude accurate positioning of the sun's 
image on a control ring at the observing 
station. Provision will be made for accu- 
rate scanning in a raster pattern over 
either the entire solar image or a small 
part of it. The data-collection system 
must collect all pertinent data and store 
them on a magnetic tape suitable for 
direct use on a large computer. Comple- 
tion is scheduled for the fall of 1964. 

The outer surface of the light shaft of 
the 150-foot solar tower telescope was 
rebuilt during the year under the super- 
vision of Couch. The original surface of 
sheet metal was in bad condition, as was 
the wooden framework to which the sheet 
metal was attached. The old metal and 
wood were removed, the inner tower was 
sealed to make it more nearly air tight, a 
new wooden framework was constructed, 
aluminum foil insulation was installed, 
and a new corrugated metal outer skin 
was attached. This outer cover was 
painted with a long-lasting paint which 
has a titanium dioxide pigment. These 
changes may be expected to improve the 
internal seeing and to make the structure 
weather-resistant . 

Other projects completed include new 
plate-baking oven controls, a device for 
directly measuring the open-loop char- 
acteristics of a 60-cps carrier servo sys- 
tem, a servo automatic recovery limit 
circuit, and the control wiring for several 
small accessory instruments. 

Prime-Focus Scanner 

The new prime-focus scanner was com- 
pleted and has been used for six months 



by Oke. The instrument has a double 
beam; one beam accepts light from the 
star plus sky, the other from the sky 
alone. A light chopper switches from one 
to the other at a frequency of 45 cps. 
The photomultiplier-tube output is fed 
into a pulse-counting system. The pulse 
counter is designed to count both up and 
down, the direction being controlled by 
the chopper on the scanner. When looking 
at star plus sky, the counter counts up; 
when looking at sky alone, it counts down. 
When the counter is properly adjusted, 
only the net star count is registered and 
the sky is completely and continuously 
eliminated. In practice the system works 
very well. Normally a star is measured 
twice at each wavelength, the star image 
being placed first in one entrance beam, 
then in the other, so that any difference 
between the two optical paths cancels 
out. After a considerable search, excel- 
lent blue and infrared photomultiplier 
tubes have been found for the scanner. 
The sensitivity of these tubes is very 
high, and the dark current is low enough 
so that the scanner becomes sky-limited 
even with a bandwidth as small as 25 A. 
Observations with a 50 A band pass have 
already been made down to 17th magni- 
tude, and it should be possible to reach 
19th or 20th magnitude in the blue and 
17th or 18th in the infrared. 

Image Tubes 

In June 1964, Ford and Baum made 
another series of experimental spectro- 
scopic observations with a two-stage RCA 
cascaded image converter in the coude 
spectrograph of the 100-inch telescope. 
The last previous test of cascaded con- 
verters at Mount Wilson had been made 
by the same workers in 1962, and sub- 
stantial improvements in these tubes have 
meanwhile been achieved through the co- 
operation of RCA with the Carnegie 
Image Tube Committee and the National 
Science Foundation. As described in Year 
Book 62, some improvements have also 
been made in the apparatus for operating 

image tubes in the coude spectrograph. 

The cascaded converters developed for 
this work have a field 40 mm in diameter 
with a resolution of about 40 line pairs 
per millimeter. A uniform parallel mag- 
netic field focuses the electronic image 
from the photocathode to an intensifying 
membrane, and similarly from the mem- 
brane to the output phosphor screen. Fol- 
lowing the phosphor screen is an F 1 
schmidt camera for photographing the 
output image with a 2.5-fold minification. 
Since the present tubes have photo- 
cathodes of the multialkali type, they 
have a useful degree of sensitivity from 
about X3500 in the ultraviolet to X7500 
in the infrared. 

More than 60 coude plates were ob- 
tained with the cascaded converter tested 
by Ford and Baum in June. They in- 
cluded stellar spectra in the region of the 
H and K lines of Ca II around X3950 and 
in the region of the weak Li I line at 
X6708. Photometric step-wedge exposures 
were also made in the same two spectral 
regions so that the gains over unaided 
photography could be measured. As pre- 
viously, the tube was operated at the 
114 - inch - focal - length coude camera, 
where it yielded minified schmidt plates 
having 7 A/mm at X3950. These plates 
were compared with unaided baked IIa-0 
plates obtained at the 32-inch-focal- 
length coude camera, where the disper- 
sion is 10 A/mm. The converter plates 
had a measured rate of blackening 8 times 
faster than the unaided plates at a density 
level around 0.6, and the granularity was 
not detectably different. Owing partly to 
imperfect performance of the schmidt 
camera behind the phosphor screen, the 
resolution on the converter plates was 
somewhat less than that on the unaided 
plates exposed in the 32-inch camera, but 
the converter images were larger. All 
factors taken into account, the gain in 
information rate at X3950 was not less 
than the 8-fold gain in blackening rate. 
At X6708 the gain over unaided Ila-F 
plates was of the same order. 

The longest exposure taken with the 



cascaded converter in June was 3.5 hours. 
The tube was not refrigerated, and the 
dark emission was low enough so that 
somewhat longer exposures, perhaps 6 or 
8 hours, would have been satisfactory. 
The 3.5-hour exposure covered the X6708 
region of the spectrum of a 9th magnitude 
GO V star in the Coma Berenices cluster, 
and the Li I line did not appear to be 

Despite the encouraging results of the 
June tests, some improvements are re- 
quired in the auxiliary optics before fur- 
ther spectroscopic observations are at- 
tempted. The tests last year of McGee's 
electronographic Lenard - window tube 
also need to be followed up with another 
round of tests before further image-tube 
apparatus for the Observatories is decided 

In February 1964, Kron, Papiashvili, 
and Breckenridge brought their elec- 
tronographic image tube from Lick Ob- 
servatory to Mount Wilson to make 
some comparative tests with the results 
that have been obtained there with 
McGee's Lenard-window tube and with 
the RCA cascaded converter. Kron's tube 
is a modification of Lallemand's classical 
system. There is a valve between the 
photocathode and the electronographic 
plates so that the plates can be put into 
the vacuum system without destroying 
the photocathode. Focusing is electro- 
static, with a 2.8-fold image minification. 

In collaboration with Wilson and 
Baum, Kron's group obtained two 10- 
exposure plates at the 114-inch-focal- 
length camera of the coude spectrograph 
at the 100-inch telescope. Each plate in- 
cluded stellar spectra in addition to 
photometric step-wedge exposures. The 
field diameter was 12 mm, and the reso- 
lution was about 40 lp/mm in the central 
half of the field. The image magnification 
was slightly larger than that for unaided 
photography at the 32-inch camera, but 
the exposure times for equal density were 
slightly longer. When all factors, includ- 
ing granularity, were taken into account, 
the gain in information rate over unaided 

photography was found to be about 2.5 

Optical Systems 

One difficulty encountered in the use of 
the image-intensifier tube in spectroscopic 
observations has been that these tubes 
with their auxiliary equipment are too 
large to be placed on the axis of a schmidt 
camera. As a replacement for the schmidt 
camera, Bo wen has investigated other 
mirror systems having a focal surface 
exterior to the system. The most promis- 
ing system appears to be one developed 
originally as a microscope objective; it 
consists of a concave and a convex mirror 
with a common center of curvature. The 
grating is placed between the two mirrors 
and reflects the light of the convex mirror 
first. The third-order spherical aberration 
becomes zero when the radius of curva- 
ture of the concave mirror is 2.618 times 
that of the convex mirror. By reducing 
this ratio slightly it is possible to intro- 
duce a small amount of aberration of the 
opposite sign to partially compensate for 
fifth and higher order aberrations and for 
the spherical aberration introduced by 
the window of the intensifier tube. Focal 
ratios as low as F 1.1 appear to be feasible 
without the use of a corrector plate. 

Mirror systems for imaging the phos- 
phor of the intensifier tube on a photo- 
graphic plate have also been studied. 
Three cases were considered, depending 
on the magnification from the phosphor 
to the plate: 

Case I. Magnification less than 0.5. 
The best solution for this case appears to 
be a standard one-mirror schmidt with a 
corrector plate figured for diverging 
rather than parallel light. The limiting 
focal ratio attainable on the image side 
is slightly lower than that of a standard 
schmidt for parallel light. 

Case II. Unit magnification. A two- 
mirror concentric system in which the 
concave mirror has twice the radius of 
curvature of the convex has some unusual 
properties. The light from the phosphor, 
which is placed on the opposite side of the 



center of curvature from the two mirrors, 
is reflected first from the concave mirror, 
then from the convex mirror, and finally 
from the concave mirror again. One of the 
valuable properties of this system is that 
a plane surface is imaged as a plane sur- 
face without curvature. Since the spheri- 
cal aberration of this system is appreci- 
able, a corrector plate is required. The 
limiting focal ratio is about 5/4 times 
that of a standard schmidt. 

Case III. Magnification greater than 2. 
A two-mirror concentric system, the light 
from the phosphor being reflected first at 
the concave and then the convex mirror, 
appears to be the best solution. In general, 
a corrector plate is required, and a focal 
ratio on the phosphor side between 1 and 
2 appears to be attainable. 

In all three mirror systems the cor- 
rector plate is placed in a diverging beam 
rather than in the parallel light of the 
normal schmidt camera. This placement 
increases appreciably the off-axis aberra- 
tions introduced by an aspherical correc- 
tor plate and suggests that a Maksutov- 
type corrector plate, whose surfaces are 
nearly concentric with the mirrors, may 
have substantial advantages. 

Galaxy Image Synthesizer 

A new instrument has been built by 
Baum for measuring the diameters of 
distant galaxies. The apparent diameters 
of galaxies belonging to a cluster provide 
an estimate of the distance of the cluster. 
The more distant the cluster, the smaller 
and fainter the member galaxies appear 
to be. The purpose of using the apparent 
diameter as a distance criterion is to pro- 
vide another test for distinguishing be- 
tween various cosmological models. In 
previous work, Baum has utilized the 
apparent magnitudes of distant galaxies 
to derive a redshift-magnitude relation. 
Diameter data obtained with the new 
instrument will, it is hoped, yield a red- 
shift-diameter relation. Owing to several 
uncertainties, including the influence of 
evolutionary changes on magnitudes, it 
will be important to observe both rela- 
tionships and to take account of both 

when choosing between different world 

This new instrument for measuring the 
diameters of galaxies is a galaxy image 
synthesizer. It is a photographic device 
that permits two plates to be exposed 
simultaneously at the prime focus of the 
200-inch telescope. One plate, with a clus- 
ter of galaxies focused upon it, is held sta- 
tionary in the focal plane of the telescope. 
The other, with a neighboring star field 
incident upon it, is oscillated in and out 
through a range of extrafocal positions. 
Each star image is thereby spread into a 
fuzzy disk whose intensity profile imitates 
that of a real E0 galaxy. A motor-driven 
cam provides the required nonsinusoidal 
plate motion. Since the real galaxies and 
the synthetic ones are photographed 
simultaneously, errors due to night-sky 
radiation, seeing, diffraction, and scatter- 
ing all cancel out. For each cluster of 
galaxies, a sequence of plate pairs is ex- 
posed with graduated amplitudes of oscil- 
lation of the moving plate. The relative 
diameters of the real galaxies will be 
judged by direct comparison with syn- 
thetic galaxies of similar magnitude, and 
a characteristic value will be assigned to 
each cluster. 

An important reason for attempting 
the synthetic-image method is that 
galaxies have no distinct boundaries; 
their outskirts taper off gradually and fade 
into the surrounding sky radiation. Con- 
sequently no unambiguous diameter can 
be recognized, except the one that char- 
acterizes the tenuous intensity profile as 
a whole. The matching of synthetic 
images with real ones circumvents this 
problem as well as avoiding the image 
disturbances already mentioned. The first 
experimental exposures were obtained by 
Baum with the galaxy image synthesizer 
at Palomar in May, but several years 
may be needed to collect a definitive 
amount of observational material. 

Diffraction Gratings 

Owing to lack of technical personnel, 
operation of the ruling engines for the 
production of diffraction gratings has 



been suspended indefinitely. This labora- 
tory has been in continuous operation 
since construction of the large "A" ruling 
engine was begun in 1912. The second, or 
"B," engine was constructed in the years 
following 1930, and a simple system for 
interferometric control was successfully 

added to it in 1962. Of the many gratings 
produced in this laboratory, some twen- 
ty-five are now in use at the Mount 
Wilson and Palomar Observatories, and 
more than twenty-five others have been 
supplied to various other observatories 
throughout the world. 


As was mentioned in the introduction, 
the Carnegie Institution has undertaken 
to investigate astronomical observing 
conditions at selected sites in the southern 
hemisphere, with special attention to 
quantitative measurement of the seeing. 
It is hoped to compare some of the best 
prospective observatory sites south of the 
equator with established observatories 
such as Mount Wilson and Palomar. 

Seeing Monitors 

Three of the portable seeing monitors 
(ASM's) referred to in Year Book 62 were 
completed in the summer of 1963, and a 
fourth in June 1964. The electronic units 
for these instruments were constructed 
by Blakee, and the completed systems 
were tested at Mount Wilson and at 
Palomar Mountain by K. Anderson and 
B. Smith. Satisfactory agreement was 
obtained between the instrumental rec- 
ords and estimates of the seeing made by 
observers at the 100- and 200-inch tele- 
scopes. Two of the ASM's were taken to 
Chile in November 1963. Another was 
operated at intervals during the winter at 
Palomar by Mr. Charles Capen and Mr. 
James Young of the Table Mountain 
Station of the Jet Propulsion Laboratory ; 
later it was taken to Australia and 
New Zealand. Regular observations were 
begun with the fourth instrument at 
Palomar in June 1964 by McCarroll. 


The interest of the University of Chile 
in advancing the science of astronomy in 
that country, and the interest of the 
Carnegie Institution in site testing, led 
to the adoption of a cooperative agree- 

ment that was signed by Rector Juan 
Gomez Millas and President Haskins on 
August 6, 1963. The cordial cooperation 
rendered by the University has been of 
the utmost importance in facilitating site 
investigations in Chile by the Institution. 

Analysis of the astronomical, meteoro- 
logical, and logistical requirements shows 
that the favored region for site investiga- 
tion in Chile lies within a distance of 100 
miles of the coastal city of La Serena and 
the nearby port of Coquimbo. It is within 
this area that the Associated Universities 
for Research in Astronomy (AURA) have 
recently purchased a tract of 180 square 
miles of mountain territory known as El 
Tot oral. This property includes some 
three or four mountain summits having 
elevations between 7000 and 9000 feet 
above sea level, which are probably 
among the most promising prospective 
observatory sites in South America. One 
of these peaks, Cerro Tololo, is being 
developed by AURA as a site for an ob- 
servatory which will have a 60-inch 
reflecting telescope. 

In November 1963, Babcock and 
Hanson, with the assistance and coopera- 
tion of Dr. J. Stock and other officials 
and personnel of the Cerro Tololo Inter- 
American Observatory, visited and in- 
spected various peaks within El Totoral, 
including Morado, Cinchado, and Pachon, 
as well as La Peineta, some 200 miles to 
the north. A station for the operation of 
one of the ASM's was established first on 
Tololo, at the invitation of AURA, and 
observations of the seeing were begun on 
December 1, 1963. 

Dr. Edward A. Ackerman, Executive 
Officer of the Carnegie Institution, to- 



gether with Babcock, visited Chile in 
February 1964 for a further inspection of 
mountain sites, with special attention to 
problems affecting Pachon. Dr. John B. 
Irwin, in June 1964, was placed in charge 
of the site investigation work in Chile. 
When problems of logistics have been 
overcome, one of the ASM's will be 
operated on Morado, at an altitude of 
7000 feet. It is hoped to establish a camp 
for operation of another ASM near the 
summit of Pachon, at 8900 feet, where 
records of wind, temperature, and hu- 
midity are now being obtained. Tempo- 
rarily, one ASM is being operated near 
the summit of Tololo. 

Other Southern Countries 

In March and April, Babcock made a 
trip to Australia and to New Zealand to 
discuss sitej testing with astronomers in 
those countries and to make a preliminary 
survey of possible sites. Through the 
cordial cooperation of Dr. Bart J. Bok, 
Director of the Mount Stromlo Observa- 
tory of the Australian National Univer- 
sity, it was possible to visit not only 
Mount Stromlo but also the new Siding 
Spring observatory station near Coona- 
barabran, some 300 miles north of Mount 
Stromlo. Babcock also made a trip to the 
northern Flinders Range, about 300 miles 
north of the city of Adelaide, to visit 
Mount Serle. 

Babcock spent some time on the South 

Island of New Zealand, at the Mount 
John University Observatory, through 
the cooperation of the astronomer in 
charge, Mr. Frank M. Bateson; he also 
made an inspection trip to Black Birch, 
a site-testing station near the north end 
of the South Island, and he viewed a 
considerable portion of the country from 
the air. 

An ASM was operated for several 
nights at Mount Stromlo and at Mount 

A great deal of valuable material relat- 
ing to earlier site surveys and to astro- 
nomical and meteorological conditions 
was supplied by the directors and per- 
sonnel of all the observatories visited. 
Officials of the European Southern Ob- 
servatory have also supplied valuable 
data relating to site-testing operations 
carried out by that organization in Africa, 
and Dr. C. Jaschek has supplied results 
of site testing in Argentina by the Ob- 
servatorio Astronomico, La Plata. As a 
result of inspection and testing so far 
carried out, and of the study of the data 
provided by other organizations, it was 
concluded that the most promising sites 
for further testing in Chile are the 
mountains known as Pachon and Morado, 
and that further site-testing work should 
be done in Australia with special refer- 
ence to the Flinders Range and with 
attention to Mount Singleton and the 
Stirling Range in the southwestern part 
of that country. 


During the report year the following 
programs have been carried out by guest 
investigators from other institutions. 

George O. Abell of the Department 
of Astronomy, University of California, 
Los Angeles, has been continuing his in- 
vestigation of rich clusters of galaxies. 
Since most of the photographic observa- 
tions were obtained in previous years, 
much of the time during the past year 
was spent on photographic reductions and 

on the setting up of photoelectric se- 
quences in the cluster fields to calibrate 
the photographic photometry of cluster 

As part of his study of the structures 
of rich clusters, Abell has been collaborat- 
ing with Professors Neyman and Scott at 
the Statistical Laboratory of the Univer- 
sity of California, Berkeley, to investigate 
the extent to which subclustering can be 
present within clusters. An important 



ramification of appreciable subclustering 
would be its effect on the interpretation 
of cluster masses obtained with the virial 
theorem. Although the work is still in 
process, preliminary results do not sug- 
gest that subclustering can substantially 
influence the masses of several clusters 
derived from the virial theorem. 

The Agena stage of the Jet Propulsion 
Laboratory Ranger 6 lunar probe was pho- 
tographed approximately one day after 
launch with the 48-inch schmidt telescope 
on January 31. The positions of the Agena 
as measured from 3 plates are being used 
for post-flight analysis by the Jet Pro- 
pulsion Laboratory. 

Dr. Lawrence H. Aller of the Univer- 
sity of California at Los Angeles and Dr. 
S. J. Czyzak of the Wright-Patterson Air 
Force Base have secured photographic 
spectrophotometric observations of a 
number of planetary nebulae with the 
100-inch coude and the B spectrograph 
attached at the Newtonian focus of the 
60-inch telescope. Emphasis was placed 

transition probabilities for forbidden lines 
of [At IV] and [S II] which fall in this 
region have been completed, and work is 
in progress on the target areas. 

Dr. V. Bumba of the Astronomical 
Institute of the Czechoslovak Academy 
of Sciences, Observatory Ondrejov, came 
to the Observatories in January 1964 
as a guest investigator supported by a 
UNESCO fellowship. His research was 
concentrated on the magnetic and ve- 
locity fields in the solar atmosphere and 
was carried out mainly in collaboration 
with Howard. It is described in more 
detail in the section on solar research. 

Dr. Philip C. Keenan of the Perkins 
Observatory obtained coude spectro- 
grams at Palomar on 5 nights and at 
Mount Wilson on 3 nights in the summers 
of 1963-1964 on two groups of red vari- 
able stars. For two out of the three SRb 
variables known to have high radial ve- 
locities (although none had been pub- 
lished for CZ Delphini), Palomar spectro- 
grams gave these results: 



Stellar Vr 


Vr (D 2 ) 

KN Aqu 


-149.3 ± 0.7 
-162.0 ± 0.5 



- 8.8 

on the photographic region of the spec- 
trum, although a number of attempts 
were made to observe some of the brighter 
nebulae in the visual region. Data for the 
photographic regions of NGC 6210 and 
II 3568 are essentially complete. Observa- 
tions on NGC 3242 have been completed, 
and the results are now being prepared 
for publication. Some observations of IC 
2003, IC 2165, IC 4634, and IC 4997 and 
of NGC 2440, NGC 4361, NGC 6543, 
NGC 6828, and NGC 6886 have been 
obtained, but additional data are needed 
before reliable relative intensities can be 
obtained for the photographic region of 
the spectrum. Theoretical calculations of 

CZ Del has the highest velocity known 
for any of these semiregular variables. 
The interstellar sodium lines have inten- 
sities consistent with the spectroscopic 
estimates of luminosity corresponding to 
the normal giant branch. B aimer emission 
was discovered in CZ Del on these plates ; 
emission appears to be a common char- 
acteristic of semiregular variables of high 
velocity, although no emission lines have 
ever been observed in the third SRb 
variable V Ursae Minoris. The survey of 
spectra of Mira variables near maximum 
light (in collaboration with Deutsch) was 
extended to seven more stars. In RV 
Aquilae, S Lacertae, RT Cygni, and W 



Lyrae, D lines were found; they are being 
measured for total absorption and radial 

Dr. Ivan R. King of the University of 
Illinois spent July 1963 tracing direct 
photographs of elliptical galaxies made 
the previous year at the Cassegrain focus 
of the 60-inch reflector. These will be 
used to determine density distributions 
with the highest possible resolution as a 
basis for dynamical studies. 

During the interval August 12-19, 
1963, Dr. K. K. Kwee of the Leiden 
Observatory used the 48-inch schmidt in 
taking photovisual plates for determining 
magnitudes of a number of RR Lyrae 
stars in a field of 6° by 6° centered around 
a = 19 h 05 m and 5 = — 18°55 / . In a previous 
investigation (Leiden Annals, 22, 1, 1962), 
photographic magnitudes of these stars 
were determined. For the intended aim, 
19 plates have been taken of this field. 
The exposure times were only 3 minutes, 
ample to reach a limiting magnitude of 16. 
By taking the plates with at least 1-hour 
intervals in time and on consecutive 
nights, it is hoped that all the plates will 
cover many different phases for each of 
the RR Lyr stars in the field. A sequence 
of comparison stars situated near the 
center of the field was previously meas- 
ured photoelectrically in two colors at 
the Lowell Observatory. Four additional 
plates were taken with the purpose of 
investigating field corrections on the 
schmidt plates. 

Dr. Kwee also made a search for ultra- 
short-period pulsating variables in five 
fields near the galactic plane in Androm- 
eda and Perseus. Ten exposures of each 
field have been made on one plate with a 
total time interval of 2 hours. During 
September and October 1963 the 20- 
inch reflector at Palomar was used in a 
program of U, B, V photometry of about 
30 short-period eclipsing binaries (periods 
from 3 to 6 days) . The observations were 
made with reference to 17 standard stars 
from Johnson's lists. For most of the 
variables about 20 measurements were 
obtained, distributed evenly over all the 

Dr. Robert H. Koch of Amherst College, 
Massachusetts, has partially completed 
light curves of U Ophiuchi with the photo- 
electric scanner at the Cassegrain focus of 
the 60-inch telescope. Seven 50-A-wide 
segments of the radiation curve of U 
Oph have been observed. Six of them 
were chosen deliberately to avoid or in- 
clude absorption lines longward of the 
Balmer discontinuity. The seventh region 
is located below the discontinuity. As U 
Oph has been fairly extensively observed 
with wide-band filters, small variations 
from the elements of the commonly 
accepted model may be expressed by fine 
structure in the light curves obtained 
with the scanner. 

Dr. W. J. Luyten of the University of 
Minnesota continued his proper-motion 
program, taking second-epoch plates with 
the 48-inch schmidt telescope. During 
five nights in November 1963, some forty 
14 by 14 inch plates were exposed, mainly 
in the regions of the Hyades and Praesepe 
clusters. A few plates were taken to con- 
tinue the experimental determination of 
parallax with this telescope. 

Dr. T. A. Matthews of the California 
Institute of Technology has continued 
the work of identifying extragalactic radio 
sources and studying the associated 
optical objects. As of the end of 1963, 82 
radio sources had been securely identified 
with extragalactic objects. It is suspected 
that this number will be substantially 
increased shortly from a study in progress 
of an additional 80 sources in the region 
from a = 21 h to 13 h , 8 > -30°. 

Among the sources already identified, 
there are seven new objects (MOO-20, 
3C43, 3C55, M03-1P, 3C109, 3C202, 
3C446) in addition to those previously 
known, or found by Sandage, which may 
be quasi-stellar sources or objects re- 
sembling them on the 48-inch plates. 

In an investigation of the optical and 
radio properties of some 42 well observed 
and well defined radio sources, Matthews 
found that the strong radio sources are 
identified with optical objects that are, 
for the most part, remarkably similar. 
They are spheroidal galaxies with lumi- 



nous elliptical-like nuclear regions and 
extensive envelopes of varying visibility. 
(They are called D galaxies on Morgan's 
system of classification.) This type of 
galaxy is often found as the brightest and 
largest member of a cluster of galaxies, 
and about 10 per cent of such galaxies 
seem to produce radio sources. Not all 
radio galaxies, however, are in clusters 
of galaxies. Other optical objects identi- 
fied with radio sources which are related 
to the D galaxies are "dumbbells" and 
"N galaxies." A dumbbell is comprised of 
two equally bright nuclear regions in a 
common extensive envelope. The N 
galaxies resemble the brilliant nuclear 
region of a D galaxy without its extensive 
envelope. All the weak radio sources, 
emitting less than ^1Q 40 ergs/sec, are 
spiral or irregular galaxies, thus being 
sharply differentiated from the spheroidal 
D galaxies and quasi-stellar objects that 
produce the strong radio sources. 

Dr. Walter E. Mitchell, Jr., of the 
Perkins Observatory, with the assistance 
of Mr. Fred Breimyer, continued his 
summer investigations of the solar spec- 
trum with the Snow telescope and Michi- 
gan spectrograph on Mount Wilson. 
Observations at the center of the sun's 
disk were made in the region XX3200-3650, 
using the spectrograph in the double-pass 
mode, for the purpose of completing a 
table of weak new features in the ultra- 
violet. With the aid of a mosaic of 16 
back-surfaced 6-inch flat mirrors, spectra 
were recorded in integrated sunlight in 
the range XX6300-6500, and spectra of 
the H and K lines as well. These records 
are comparable in quality with those 
made with the telescope in the con- 
ventional way. On highly compressed 
records of this kind, the diffuse, shallow 
features at XX6263, 6341, and 6318 were 
recognized as arising from the Ca I atom 
in autoionization. A curve of growth for 
the water vapor features near 1 micron 
was derived from sec z observations of the 
sun in that region. Surface humidity data 
were recorded during the period of the 
observations with a view to establishing 
the correlation between surface humidity 

and the precipitable water vapor that 
may be spectroscopically estimated to 
lie in the entire air path. Experimentation 
continued with the ratio recording of a 
signal and monitor beam from the 
spectrograph. On one occasion compensa- 
tion was successful to the extent that 
variations of 5 to 10 in sky transparency 
could be nullified. In a program being 
conducted with Howard on the rms 
magnetic field of the sun, the profiles of 
18 lines of moderate strength and different 
magnetic splitting were recorded at the 
center of the solar disk. 

Dr. B. C. Murray, Dr. R. L. Wildey, 
and Mr. J. A. Westphal of the Division of 
Geological Sciences of the California 
Institute of Technology have continued 
their collaboration in the use of the 200- 
inch telescope for photometry at 10 
microns. Brightness temperature maps of 
Venus have now been produced with full 
phase coverage; they appear to be in- 
consistent with limb darkening super- 
imposed on a simple latitude-dependent 
morphology. Mr. Westphal has continued 
the synoptic infrared observations of 
Venus throughout the spring. Jupiter has 
been reobserved with superior technique 
that provides doubled angular resolution 
and improvement in the signal-to-noise 
ratio by a factor of 4. Significant correla- 
tion of band structure and thermal emis- 
sion is shown by the new observations, 
and it is found that the Great Red Spot 
has a much lower brightness temperature 
than its surroundings. The eclipse of J m 
by Jupiter was successfully observed. The 
extremely rapid cooling and reheating 
implies that the outer layer, to a depth 
of at least a millimeter, has insulating 
properties at least as effective as those of 
the outer layer of the moon. 

Dr. Wildey, using the 200-inch tele- 
scope, has extended the photometry 
of stars at 10 microns to include 30 ad- 
ditional stars. 

Dr. Wildey and Mr. H. A. Pohn have 
concluded a program of detailed photo- 
electric photometry of the moon. No 
phase lag in maximum brightness was 
found. Using simultaneously the U, B, V 



photoelectric photometer and a photo- 
metric camera attached to the 20-inch 
telescope at Palomar, the same investiga- 
tors began work on a photographic atlas 
of the moon with photoelectric calibra- 
tions. Wildey used the coude scanner of 
the 100-inch telescope to measure differ- 
ential luminescence on the moon; he 
found the amount less than 2 per cent of 
the level of the continuous spectrum at 
X3900. He has also developed a computer 
program for finding selenographic coor- 
dinates of infrared observations on the 
moon's dark side, using standard lunar 
reference features on the bright side, 
accurate differential measurements of 
right ascension and declination, and 
ephemeris boundary values. After theo- 
retical investigation of the implications 
of the enhancement of thermal emission 
from Jupiter's satellite shadows, it is 
concluded that explanations in terms of 
alteration of physical equilibrium, such as 
change of state, seem untenable; a 
photochemical process appears to be 

The program of photoelectric spectro- 
photometry of planetary nebulae has been 
continued by Dr. Charles R. O'Dell of 
the University of California to include all 
bright nebulae, with particular emphasis 
on hydrogen and helium series, to deter- 
mine more accurately the intrinsic varia- 
tions of low members. Deviations from 
existing recombination theory definitely 
occur, many of which can be explained as 
being due to partial collisional excitation 
and self-absorption from the 2S state of 
hydrogen; the very flat Balmer decre- 
ment for low members, however, re- 
mains unexplained. Measures were also 
made of NGC 6644 and IC 4846, both 
planetary nebulae lying in the direction of 
the galactic nucleus and having large 
radial velocities (+194 and +151 km/ 
sec, respectively). The abundance ratio 
of helium and hydrogen was found to be 
N(Re)/N(R) = 0.14 ± 0.03, a value 
very similar to values for field planetary 
nebulae of much different kinematic 
properties. Because prior photoelectric 

studies of the helium line intensities in 
the Orion nebula are in poor agreement, a 
central region of M 42 was measured to 
obtain the line-intensity ratios Ha/H/3/ 
H7/H5/X5876/X4471. After comparison 
with other studies and correction for 
interstellar reddening, an abundance 
ratio of N(He)/N(R) = 0.14 ± 0.02 was 
derived — a larger value than has been 
given before in the literature on this 

Dr. Bernard Pagel of the Royal Green- 
wich Observatory, working with the 100- 
inch telescope, secured 18 coude plates of 
7 Virginis A and B, HD 71377, x Herculis, 
45 Bootis, and 5 Serpentis A in the blue, 
yellow, and red regions, and 6 sky plates. 
These are being traced at Herstmonceux 
with a view to abundance determinations 
in connection with the following prob- 
lems: effect of rotation on analysis, homo- 
geneity of stellar groups, abundance 
anomalies in mild subdwarfs hotter than 
the sun, and calibration of abundance 
estimates based on six-color photometry. 
Pagel also obtained microphotometer 
tracings of ten 200-inch coude plates of 
Ai Cassiopeiae and HD 25329 supplied by 
Greenstein; they will be used for detailed 
abundance studies of these two sub- 

Dr. Daniel M. Popper of the University 
of California at Los Angeles has pro- 
gressed with spectrographic observations 
of eclipsing binaries at Mount Wilson. 
This program is aimed at more precise 
knowledge of the masses of stars of 
different types. It was found that MY 
Cygni is a new metallic-lined eclipsing 
binary with double lines, so that an addi- 
tional mass determination should be 
possible for this class. 

Six of the so-called Trumpler stars were 
observed with the Mount Wilson tele- 
scopes by Dr. Jorge Sahade of the 
Observatorio Astronomico, La Plata, 
Argentina, with the aim of explaining the 
redshift that had been interpreted as 
indicative of large masses. A few plates of 
the Be star HD 20336 and of HD 698 and 
17 Leporis were also secured. 



Photoelectric U, B, V magnitudes in 
the range V= 10 m to l?^ for stars in the 
Small Sagittarius Cloud, in NGC 6871 
and 6883, and in M 37 have been obtained 
by Tammann, working as a guest investi- 
gator with the 60- and 100-inch. These 
stars will be used as standards for the 
Basel Observatory program of a three- 
color photometric investigation on 48- 
inch plates of star fields in different 
galactic latitudes and longitudes as well 
as of some galactic clusters. The 20-inch 
was used in combination with the RCA 
7265 phototube to get the first photo- 
electric magnitudes in W. Becker's R, G, 
U system. This system is photographi- 
cally defined, and the best approximation 
for photoelectric work was obtained with 
the filters Schott RG1, GG5+BG7, and 
UG2, respectively. A fourth color with 
the filters GR1+UG2 was measured only 
to correct for the red leak of the ultra- 
violet filter. 

With the nebular spectrograph at the 
60-inch, Tammann obtained some spectra 
of four bright cluster members of NGC 
7790 as well as of some comparison stars. 
These spectra will be used to determine 
the radial velocity of the cluster, and with 
this to obtain an additional membership 
criterion for the double cepheid CE 
Cassiopeiae, the cepheid CF Cass, and 
the eclipsing binary QX Cass. 

Radiometric and photometric mapping 
of the moon through a lunation was 
carried out with the 60-inch telescope by 
Dr. Richard W. Shorthill and Dr. John 
M. Saari of the Boeing Scientific Re- 
search Laboratories. For this work a 
focal-plane scanner was mounted at the 
Newtonian focus. With a separation be- 
tween scan lines equal to the diameter of 
the scanning aperture (8"), a scan of the 
full moon requiring a raster of 240 lines 
could be made in less than 30 minutes. 
(Auxiliary photographs were obtained 
with the 60-foot solar tower and an 8-inch 
telescope mounted on the 60-inch.) The 
illuminated lunar disk was simultaneously 
mapped with an infrared detector (10 to 
12 y. band pass) and a photomultiplier 

(4450 A peak response). The purpose of 
the program was to produce isothermal 
and isophotic contours that could be re- 
lated to visible surface features. In six 
periods of observing between July and 
December 1963, mapping of the moon at 
35 different phases was completed. 
Measurements could not be made during 
the December 30 lunar eclipse because of 
a haze. 

As part of a program to determine the 
reddening and distance to certain CH 
stars, Dr. George Wallerstein of the 
University of California used the X 
spectrograph on the 60-inch to obtain 
classification spectrograms of selected A 
and B stars. 

Dr. Ray Weymann of the Steward 
Observatory, University of Arizona, using 
the 100-inch telescope, obtained spectro- 
grams at 4.5 A/mm showing emission 
components of the H and K lines in G, K, 
and M stars of all luminosity classes. 
Observations of the infrared triplet of 
Ca II at 6 A/mm failed to reveal any sign 
of analogous emission. Plates well suited 
for microphotometering the emission 
profiles of approximately 40 stars have 
now been obtained at the Mount Wilson 
and Lick Observatories. These data are 
now being reduced, and the relevant 
parameters describing the profiles will be 
published shortly. 

A study is being made by Drs. G. O. 
Abell, G. E. Kocher, and A. G. Wilson of 
the Rand Corporation to determine the 
limiting magnitude of the schmidt tele- 
scopes when photographing moving ob- 
jects under sky brightness conditions 
varying from dark to full moon. It is 
desired to find the optimum emulsion and 
filter combination for use in moonlight, 
as well as the limiting magnitude as a 
function of the rate of angular motion of 
the object. 

A. G. Wilson and G. E. Kocher are con- 
tinuing investigations of comparative 
galaxy diameters using, among other 
techniques, iterative reproductions on 
high-contrast emulsions. Transfer plates 
between selected galaxies and clusters of 



galaxies are being taken to secure com- 
parable images. The principal purpose of 
the program is to investigate possible size 
discretization relations and to test the 
Edelen \/n(n-\-l) discretization hypoth- 
esis for elliptical galaxies. 

Dr. R. v. d. R. Woolley of the Royal 
Greenwich Observatory took a number of 
direct photographs of NGC 6522 besides 
continuing determinations of radial veloc- 
ity with the coude spectrograph. The 
direct photographs were repeat plates of 
exposures made by the late Walter 
Baade. The purpose of taking the repeat 

plates was twofold: to attempt to de- 
termine the proper motion of the cluster, 
and to attempt to find at least some infor- 
mation about the velocity dispersion 
of the field stars at the center of the 
Galaxy. Two pairs of plates, one of each 
epoch, have been measured at Herstmon- 
ceux, and the results have been closely 
studied. The time interval is a little short 
for the purpose, and the object is in- 
conveniently far to the south, but it is 
hoped to arrive at some tentative con- 
clusions shortly. 


The retirement of Dr. Ira S. Bowen 
after 18 years as Director, to become 
Distinguished Service Staff Member, was 
noted in the introduction. Horace W. 
Babcock was appointed to succeed him as 
Director, effective July 1, 1964. 

Other changes in the organization that 
have occurred during the year include the 

Dr. Robert B. Leighton, a member of 
the Observatory Committee and Profes- 
sor of Physics at the California Institute 
of Technology, was appointed to the 
staff of the Observatories as of July 1, 
1963. For several years Dr. Leighton's 
chief research interests have been in the 

field of solar physics, and he is also 
concerned with many problems of astro- 
nomical instrumentation. 

Dr. Edwin W. Dennison, whose special 
interests lie in electronic instrumentation, 
came from the Sacramento Peak Observa- 
tory to accept an appointment as Staff 
Member, effective September 1, 1963. 

Dr. John B. Irwin took up his appoint- 
ment as Staff Associate on June 6, 1964, 
with responsibility for site-testing opera- 
tions in Chile. 

Mrs. Mary F. Coffeen retired after five 
years as Librarian and many years as a 
research assistant. She first came to the 
Observatory as a computer in 1922. 

Research Division 

Staff Members 

Halton C. Arp 

Horace W. Babcock, Associate Director 

William A. Baum 

Ira S. Bowen, Director 

Edwin W. Dennison 

Armin J. Deutsch 

Olin J. Eggen 

Jesse L. Greenstein 

Robert F. Howard 

Robert P. Kraft 

Robert B. Leighton 

Guido Munch 

J. Beverley Oke 

Allan R. Sandage 

Maarten Schmidt 

Olin C. Wilson 

Fritz Zwicky 

Staff Members Engaged in Post-Retirement 

Harold D. Babcock 
Alfred H. Joy 

Senior Research Fellows 

Leonard T. Searle 1 
Arne A. Wyller 2 

Carnegie Research Fellows 

Leonard V. Kuhi 
Hugo van Woerden 
John B. Whiteoak 

Research Fellows 
Claude Arpigny 3 

1 Resigned July 31, 1963. 

2 Resigned June 30, 1964. 

3 Resigned March 31, 1964. 



Bodo Baschek 4 
Jacques Berger 5 
Eugene Capriotti 4 
Peter S. Conti 
Ivan J. Danziger 
Rolf P. Fenkart 
John E. Gaustad 
Robert Kovar 
Antoni Stawikowski 
Henrietta H. Swope 
Robert L. Wildey 

Senior Research Assistant 
Dorothy D. Locanthi 

Research Assistants 

Christine Arpigny 6 

Jeanne Berger 7 

Frank J. Brueckel 

Sylvia Burd 

Rowena Danziger 

Mary F. Coffeen, Librarian 8 

Thomas A. Cragg 

Emil Herzog 

Gertrud Herzog 

Maria Karpowicz 

Basil Katem 

Margaret Katz 

Charles T. Kowal 

A. Louise Lowen 

George P. Mylonas 

Joyce E. Sheeley 9 

Gustav A. Tammann 

Merwyn G. Utter 

Mary B. Whiteoak 10 

Student Observers 
Subhash Chandra 
Theodore N. Divine 
James E. Gunn 
Tom Kennedy Greenfield 
Manuel E. Mendez 
Robert H. Norton 
Alan M. Title 
Laurence M. Trafton 
Neil R. Sheeley 

William C. Miller 

Ester Bru Baum 

4 Resigned September 30, 1963. 

5 Resigned October 1, 1963. 

6 Resigned March 20, 1964. 

7 Resigned October 11, 1963. 

8 Retired June 30, 1964. 

9 Resigned June 12, 1964. 

10 Resigned February 17, 1964. 

Instrument Design and Construction 

Lawrence E. Blake6, Senior Electronic 

Eileen I. Challacombe, Draftsman 

Floyd E. Day, Optician 

Kenneth E. DeHuff, Machinist 

Robert D. Georgen, Machinist 

Melvin W. Johnson, Optician 

Rudolf E. Ribbens, Designer and Super- 
intendent of Instrument Shop 

Stuart L. Roberts, Instrument Maker 11 

Bruce Rule, Project Engineer 

Marlin N. Schuetz, Electronic Technician 

John Shirley, Electronic Engineer 

Benny W. Smith, Electronic Technician 

Maintenance and Operation 
Mount Wilson Observatory Offices 

Paul F. Barnhart, Chauffeur 

Wilma J. Berkebile, Secretary 

Hugh T. Couch, Superintendent of Build- 
ings and Grounds 

Helen S. Czaplicki, Typist Editor 

Fannie G. Gabrielsen, Stewardess 

Eugene L. Hancock, Night Assistant 

Mark D. Henderson, Gardener 

Anne McConnell, Administrative Assistant 

Leah M. Mutschler, Stenographer and 
Telephone Operator 

Bula H. Nation, Head Stewardess 

Alfred H. Olmstead, Night Assistant 

Arnold T. Ratzlaff, Night Assistant 12 

Glen Sanger, Custodian 

Henry P. Schaefer, Night Assistant 

John E. Shirey, Laborer 

William D. St. John, Custodian and Relief 

Benjamin B. Traxler, Superintendent 

Palomar Observatory and Robinson Laboratory 

Fred Anderson, Machinist 

Jan A. Bruinsma, Custodian 

Maria J. Bruinsma, Stewardess 

Eleanor G. Ellison, Secretary and Librarian 

Leslie S. Grant, Relief Night Assistant and 

Mechanic 13 
Victor A. Hett, Night Assistant 
Byron Hill, Superintendent 
Helen D. Hollo way, Secretary 
Charles E. Kearns, Night Assistant 

11 Resigned May 30, 1964. 

12 Retired (disability) October 1, 1963. 

13 Resigned May 1, 1964. 



J. Luz Lara, Ground Mechanic 
Harley C. Marshall, Office Manager 14 
D wight M. Miller, Mechanic 15 
Robert D. Quinn, Maintenance Mechanic 
Gary M. Tuton, Night Assistant 
Hendrika E. van Buuren, Stewardess 
John E. van Buuren, Custodian 
William C. Van Hook, Electrician and 
Assistant Superintendent 

14 Resigned September 29, 1963. 
16 Resigned July 18, 1963. 

Betty A. Wallace, Secretary 
Gus Weber, Assistant Mechanic 

Site-Testing Operations 

James N. Hanson, Staff Associate 3 
John B. Irwin, Staff Associate 
Peter Konrad, Assistant Observer 
Manfred Wagner, Observer 

Temporary Research Assistants 

Kurt Anderson 
David McCarroll 


Abell, G. O., see Minkowski, R. L. 

Aller, L. H., I. S. Bowen, and 0. C. Wilson, The 
spectrum of NGC 7027, Astrophys. J., 188, 
1013-1017, 1963. 

Arp, Halton C., Spiral structure in M 31, 
Astrophys. J., 139, 1044-1057, 1964. 

Arp, Halton C., A technique for faint photogra- 
phy applied to the radio source in Fornax, 
Astrophijs. J., 189, 1378-1380, 1964. 

Arp, Halton C., The role of the Magellanic 
Clouds in the understanding of the galaxies, 
The Galaxy and the Magellanic Clouds, Intern. 
Astron. Union-Union Radio Sci. Intern. Symp. 
20, 219-227, edited by F. J. Kerr and A. W. 
Rodgers, Australian Academy of Science, Can- 
berra, 1964. 

Arp, Halton C, see also Baade, Walter. 

Arpigny, Claude, Approximate evaluation of the 
Fermi-Dirac functions, Astrophys. J., 138, 
607-609.. 1963. 

Baade, Walter, and Henrietta H. Swope, Vari- 
able star field 96' south preceding the nucleus 
of the Andromeda Galaxy, Astroyi. J., 68, 435- 
469, 1963. 

Baade, Walter, and Halton Arp, Positions of 
emission nebulae in M 31, Astrophys, J., 189, 
1027-1044, 1964. 

Babcock, Horace W., The sun's magnetic 
field, Ann. Rev. Astron. Astrophys., 1, 41-58, 
Annual Reviews, Palo Alto, California, 1963. 

Baschek, Bodo, The near infrared spectrum of 
Nova Herculis 1963, Publ. Astron. Soc. Pacific, 
76, 22-27, 1964. 

Berger, Jacques, Radial velocities and spectra of 
faint blue stars at high galactic latitudes, 
Publ. Astron. Soc. Pacific, 75, 393-399, 1963. 

Berger, Jacques, and Jesse L. Greenstein, A new 
helium-rich star, BD + 13°3224, Publ. Astron. 
Soc. Pacific, 75, 336-342, 1963. 

Berger, Jacques, see also Zwicky, Fritz. 

Bowen, Ira S., see Aller, L. H. 

Burbidge, E. M., G. R. Burbidge, and A. R. 
Sandage, Evidence for violent events in the 
nuclei of galaxies, Rev. Mod. Phys., 85, 947- 
972, 1963. 

Burbidge, G. R., see Burbidge, E. M. 

Cayrel, Roger, and Jun Jugaku, Predicted fluxes 
and line intensities in late-type model atmos- 
pheres, Ann. Astrophys., 26, 495-560, 1963. 

Cragg, Thomas, R. Howard, and H. Zirin, Verti- 
cal structures in the atmosphere, Astrophys. 
J., 138, 303-305, 1963. 

Deutsch, Armin J., D-line intensities in the 
spectra of galaxies, Astrophys. J., 189, 532- 
538, 1964. 

Eggen, Olin J., Three-color photometry of the 
components in 228 wide double and multiple 
systems, Astron. J., 68, 483-514, 1963. 

Eggen, Olin J., Luminosities, colors, and motions 
of the brightest A-type stars, Astron. J., 68, 
697-714, 1963. 

Eggen, Olin J., Color-luminosity array for NGC 
752, Astrophys. J., 138, 356-365, 1963. 

Eggen, Olin J., Ages and kinematics of clusters, 
The Galaxy and the Magellanic Clouds, Intern. 
Astron. Union-Union Radio Sci. Intern. 
Symp. 20, 10-15, edited by F. J. Kerr and 
A. W. Rodgers, Australian Academy of 
Science, Canberra, 1964. 

Eggen, Olin J., The angular momentum in visual 
binaries, Publ. Astron. Soc. Pacific, 75, 308- 
313, 1963. 

Eggen, Olin J., Clusters and stellar evolution: 
a colloquium sponsored by Commission 37 of 
the I. A. U., edited by O.J. Eggen and G. H. 
Herbig, Roy. Obs. Bull. 82, ser. E, 108 pp., 

Eggen, Olin J., A catalogue of high velocity 
stars, Roy. Obs. Bull. 84, ser. E, 203 pp., 1964. 

Gaposchkin, Cecilia Payne, The 1960 minimum 
of R Coronae Borealis, Astrophys. J., 138, 320- 
341, 1963. 



Gates, H. S., see Zwicky, Fritz. 

Gratton, L., see Sandage, Allan. 

Greenstein, Jesse L., Astronomy, Collier's 1963 
Yearbook, pp. 91-93, Crowell-Collier Publish- 
ing Company, New York. 

Greenstein, Jesse L., The white dwarf L770-3, 
Harvard Announcement Card 1644, Harvard 
Observatory, April 21, 1964. 

Greenstein, Jesse L., Solar and stellar magnet- 
ism, Science, 142, 686, 1963. 

Greenstein, Jesse L., Zvedzii Atmospharii, edited 
by J. L. Greenstein, Russian translation by 
V. V. Sobolev of vol. VI of Stars and Stellar 
Systems, Moscow, 1963. 

Greenstein, Jesse L., see also Berger, Jacques; 
Heifer, H. L.; Keenan, Philip C.; Wallerstein, 

Griffin, Roger F., The positions of optical objects 
in the fields of 42 radio sources, Astron. J., 68, 
421-428, 1963. 

Harvey, J. W., see Howard, Robert. 

Hazlehurst, John, The composition of 5 Eridani, 
Observatory, 83, 128-133, 1963. 

Heifer, H. L., G. Wallerstein, and J. L. Green- 
stein, Metal abundances in the subgiant Zeta 
Herculis and three other dG stars, Astrophys. 
J., 138, 97-117, 1963. 

Herzog, E., see Zwicky, Fritz. 

Howard, Robert, On the relation of major solar 
flares with changes in sunspot areas, Astrophys. 
J., 138, 1312-1313, 1963. 

Howard, Robert, and J. W. Harvey, Photo- 
spheric magnetic fields and chromospheric 
features, Astrophys. J., 139, 1328-1335, 1964. 

Howard, Robert, see also Cragg, Thomas. 

Humason, M. L., see Zwicky, Fritz. 

Joy, Alfred H., The beginnings of the Astro- 
nomical Society of the Pacific, Publ. Astron. 
Soc. Pacific, 76, 1-5, 1964. 

Jugaku, Jun, and W. L. W. Sargent, Ultraviolet 
spectrum of 3 Centauri A, Astrophys. J., 138, 
90-96, 1963. 

Jugaku, Jun, see also Cayrel, Roger. 

Kaplan, Lewis D., G. Munch, and Hyron Spin- 
rad, An analysis of the spectrum of Mars, 
Astrophys. J., 139, 1-15, 1964. 

Katem, Basil, see Sandage, Allan. 

Keenan, Philip C., and Jesse L. Greenstein, The 
line spectrum of R Coronae Borealis, XX3700- 
8600 A, published by Ohio State University 
Engineering Experiment Station, Columbus, 
pp. 199-231, 1963. 

Kowal, Charles, see Sandage, Allan. 

Kraft, Robert P., Cataclysmic variables as bi- 
nary stars, Advan. Astron. Astrophys., 2, 43- 
85, edited by Z. Kopal, Academic Press, New 
York and London, 1963. 

Kraft, Robert P., Are all novae binary stars? 
Astron. Soc. Pacific Leaflet 418, 8 pp., April 

Kraft, Robert P., Binary stars among cata- 
clysmic variables, III, Ten old novae, Astro- 
phys. J., 139, 457-475, 1964. 

Kraft, Robert P., The absolute magnitudes of 
classical cepheids, Stars and Stellar Systems, 
vol. Ill, Basic Astronomical Data, chapter 21, 
pp. 421-447, edited by K. Aa. Strand, Univer- 
sity of Chicago Press, 1963. 

Kraft, Robert P., and Maarten Schmidt, Galactic 
structure and rotation from cepheids, The 
Galaxy and the Magellanic Clouds, Intern. 
Astron. Union-Union Radio Sci. Intern. 
Symp. 20, 68-69, edited by F. J. Kerr and 
A. W. Rodgers, Australian Academy of 
Science, Canberra, 1964. 

Matthews, Thomas A., and Allan Sandage, 
Optical identifications of 3C48, 3C196, and 
3C286 with stellar objects, Astrophys. J., 138, 
30-56, 1963. 

Matthews, Thomas A., see also Schmidt, 

Mihalas, Dimitri, Model atmosphere studies of 
early-type stars (abstract), Astron. J., 69, 
144-145, 1964. 

Mihalas, Dimitri, Photoelectric equivalent 
widths in p Leo and -n- 4 Ori, Astrophys. J., 139, 
764-765, 1964. 

Mihalas, Dimitri, Decay times of type I super- 
novae light curves, Publ. Astron. Soc. Pacific, 
75, 256-268, 1963. 

Miller, William C, see Sandage, Allan R. 

Minkowski, R. L., and G. O. Abell, The National 
Geographic Society-Palomar Observatory 
Sky Survey, in Stars and Stellar Systems, vol. 
Ill, Basic Astronomical Data, appendix II, pp. 
481-487, edited by K. Aa. Strand, University 
of Chicago Press, 1963. 

Munch, Guido, Review of ''Interstellar Matter 
in Galaxies," 330 pp., edited by L. Woltzer, 
W. A. Benjamin, Inc., New York, 1962, Am. 
Scientist, 51, 440A, 1963. 

Munch, Guido, see also Kaplan, Lewis D. 

Murray, Bruce C, Robert L. Wildey, and James 
A. Westphal, Observations of Jupiter and the 
Galilean satellites at 10 microns, Astrophys. 
J., 139, 986-993, 1964. 

Murray, Bruce C, see also Wildey, Robert L. 

O'Dell, C. R., The evolution of the central stars 
of planetary nebulae, Astrophys. J., 138, 67- 
78, 1963. 

O'Dell, C. R., Photoelectric photometry of the 
planetary nebulae, III, Astrophys. J., 138, 
293-294, 1963. 



O'Dell, C. R., Photoelectric spectrophotometry 
of planetary nebulae, Astrophys. J., 138, 1018- 
1034, 1963. 

Parker, Robert A. R., Physical conditions in the 
Cygnus Loop and some other possible super- 
nova remnants, Astrophys. J., 139, 493-512, 

Rudnicki, Konrad, Indicated intergalactic ab- 
sorption in regions of the sky occupied by the 
largest near and medium distant clusters listed 
in the Catalogue of Galaxies and Clusters of 
Galaxies by Zwicky, Herzog, and Wild, Acta 
Astron., 13, 165-168, 1963. 

Rudnicki, Konrad, The visible distribution of 
galaxies in the Perseus cluster, Acta Astron., 
13, 230-242, 1963. 

Rudnicki, Konrad, see also Zwicky, Fritz. 

Ryle, M., and Allan Sandage, The optical identi- 
fication of three new radio objects of the 3C48 
class, Astrophys. J., 139, 419-421, 1964. 

Sandage, Allan, Photoelectric observations of the 
interacting galaxies VV 117 and VV 123 re- 
lated to the time of formation of their satel- 
lites, Astrophys. J., 138, 863-872, 1963. 

Sandage, Allan, Intensity variations of 3C48, 
3C196, and 3C273 in optical wavelengths, 
Astrophys. J., 139, 416-419, 1964. 

Sandage, Allan, Results of a pilot program to 
discover new subdwarfs in the solar neighbor- 
hood, Astrophys. J., 139, 442-450, 1964. 

Sandage, Allan, The cosmological problem, J. 
Quant. Spectr. Radiative Transfer, 3, 541-549, 

Sandage, Allan, and L. Gratton, Observational 
approach to stellar evolution, Proc. Intern. 
School of Physics, Enrico Fermi, Course 
XXVIII, Star Evolution, pp. 11-49, edited by 
L. Gratton, Academic Press, New York and 
London, 1963. 

Sandage, Allan, and Basil Katem, Three-color 
photometry of the metal-rich globular cluster 
NGC 6171, Astrophys. J., 139, 1088-1094, 

Sandage, Allan, and Charles Kowal, Optical de- 
tection of the West Ford belt at Palomar, 
Science, 141, 797-798, 1963. 

Sandage, Allan, and William C. Miller, The 
exploding galaxy M82 : Evidence for the exist- 
ence of a large-scale magnetic field, Science, 
144, 405-409, 1964. 

Sandage, Allan, see also Burbidge, E. M.; 
Matthews, Thomas A.; Ryle, M. 

Sargent, Wallace L. W., Leonard Searle, and 
George Wallerstein, On a recent abundance 
analysis of 7 Sextantis, Astrophys. J., 139, 
1015-1017, 1964. 

Sargent, Wallace L. W., see also Jugaku, Jun; 
Searle,' Leonard. 

Schatzman, E., On the acceleration of particles 
in shock fronts, Ann. Astrophys., 26, 234-249, 

Schmidt, Maarten, and Thomas A. Matthews, 
Redshifts of the quasi-stellar radio sources 
3C47 and 3C147, Astrophys. J., 139, 781-785, 

Schmidt, Maarten, see also Kraft, Robert P. 

Searle, Leonard, and Wallace L. W. Sargent, 
Studies of the peculiar A stars, II, The silicon- 
abundance anomaly, Astrophys. J., 139, 793- 
812, 1964. 

Searle, Leonard, see also Sargent, Wallace L. W. 

Spinrad, Hyron, see Kaplan, Lewis D. 

Swope, Henrietta H., Note on planetary nebulae 
in M31, Astron. J., 68, 470, 1963. 

Swope, Henrietta H., see also Baade, Walter. 

Tammann, Gustav Andreas, Die explodierende 
Galaxie M82, Sterne und Weltraum, 3, 9-12, 
1964; Umschau, 64, 229-232, 1964. 

Utter, Merwyn G., The heavens in 1964, Astron. 
Soc. Pacific, annual series, 8 pp., January 1964. 

Wallerstein, George, and Jesse L. Greenstein, 
The chemical composition of two CH stars, 
HD 26 and HD 201626, Astrophys. J., 139, 
1163-1179, 1964. 

Wallerstein, George, see also Heifer, H. L.; 
Sargent, Wallace L. W. 

Westphal, James A., see Murray, Bruce C. 

Wildey, Robert L., Photoelectric and photo- 
graphic observations of 944 stars in and near 
h and x Persei (abstract), Astron. J., 68, 546, 

Wildey, Robert L., The stellar content of h and 
X Persei — cluster and association, Astrophys. 
J., Suppl. Ser. 84, 439-496, 1964. 

Wildey, Robert L., Lunar luminescence, Publ. 
Astron. Soc. Pacific, 76, 112-114, 1964. 

Wildey, Robert L., and Bruce C. Murray, 10-/i 
photometry of 25 stars from B8 to M7, 
Astrophys. J., 139, 435-441, 1964. 

Wildey, Robert L., and Bruce C. Murray, Ten- 
micron stellar photometry — First results 
and future prospects, Colloque-Inter national 
d' Astrophysique tenu a V University de LiCge, 
IX, 460, 1964. 

Wildey, Robert L., see also Murray, Bruce C. 

Wilson, Olin C., A probable correlation between 
chromospheric activity and age in main- 
sequence stars, Astrophys. J., 138, 832-848, 
Wilson, Olin C., Paul Willard Merrill, 1887-1961, 
Biographic Memoirs, 37, 237-266, published 
for National Academy of Sciences of the 
United States by Columbia University Press, 



Wilson, Olin C, Review of "Evolution of Stars 
and Galaxies" by Walter Baade, edited by 
Cecilia Payne-Gaposchkin, Harvard Univer- 
sity Press, Cambridge, Massachusetts, 1962, 
Publ. Astron. Soc. Pacific, 75, 202-203, 1963. 

Wilson, Olin C, The distribution of intensities 
of bright H and K in dK stars and the rate of 
star production in the Galaxy, Publ. Astron. 
Soc. Pacific, 76, 28-34, 1964. 

Wilson, Olin C, see also Aller, L. H. 

Zirin, H., see Cragg, Thomas. 

Zwicky, Fritz, l'Univers vu par les astronomes, 
Ann. Guibhard, 39, 263-274, 1963. 

Zwicky, Fritz, NGC 1058 and its Supernova 
1961, Astrophys. J., 139, 514-519, 1964. 

Zwicky, Fritz, Polarization of M 82 by com- 
posite analytical photography, Astrophys. J., 
139, 1394-1395, 1964. 

Zwicky, Fritz, List of supernovae discovered 
since 1885, California Institute of Technology, 
Pasadena, June 1964. 

Zwicky, Fritz, A field rich in clusters of galaxies, 
Publ. Astron. Soc. Pacific, 75, 373-375, 1963. 

Zwicky. Fritz, New types of celestial objects, 
Sky and Telescope, 26, 83, 1963. 

Zwicky, Fritz, J. Berger, H. S. Gates, and K. 
Rudnicki, The 1962 Palomar supernova 
search, Publ. Astron. Soc. Pacific, 75, 236-238, 

Zwicky, Fritz, and E. Herzog, Catalogue of 
galaxies and of clusters of galaxies, vol. II, 
371 pp., California Institute of Technology, 
Pasadena, 1963. 

Zwicky, Fritz, and M. L. Humason, Spectra and 
other characteristics of interconnected galaxies 
and of galaxies in groups and in clusters, III, 
Astrophys. J., 139, 269-283, 1964. 

Zwicky, Fritz, and M. L. Humason, Spectra and 
other characteristics of interconnected galaxies 
and of galaxies in groups and in clusters, III, 
a correction, Astrophys. J., 139, 1393-1394, 

Geophysical Laboratory 

Philip H. Abelson 

Director Washington, District of Columbia 

Carnegie Institution of Washington Year Book 63, 1963-1964 


Introduction 59 

Experimental Petrology 64 

An evaluation of the new progress . 64 

Crystal and liquid trends in simplified 

alkali basalts 65 

The peralkaline residua system : 

Na 2 0-Al 2 03-Fe 2 03-Si0 2 .... 74 
Ijolite and nephelinite as residual 

liquids 75 

Transitions from oversaturated to 

undersaturated compositions . . 76 
Peralkaline liquids and the norm . . 78 

Sodium silicates 78 

Temperature and vapor composition in 

carbonatite and kimberlite ... 79 
Breakdown of monticellite and akerman- 

ite at high pressures 81 

Wollastonite-pseudowollastonite inversion 83 
Stability field of akermanite .... 84 

Soda melilite 86 

The join akermanite (Ca 2 MgSi 2 07)-soda 

melilite (NaCaAlSi 2 7 ) .... 89 
The join akermanite-soda melilite at 20 

kilobars 90 

New relations on melting of basalts . . 92 
Genesis of principal basalt magmas . . 97 
The system diopside-forsterite-enstatite 

at 20 kilobars 101 

The join diopside-forsterite at 20 

kilobars 101 

The join wollastonite-enstatite at 20 

kilobars 103 

The join wollastonite-diopside . . 103 
The join diopside-enstatite . . .103 
The system diopside-forsterite-ensta- 
tite at 20 kilobars 105 

Experimental studies on the basalt- 

eclogite transformation . . . .108 
Pressure-temperature plane for anor- 

thite + forsterite composition . . 109 
Pressure-temperature plane for anor- 

thite + 2 enstatite composition . 112 
Pyroxene fractionation in mafic magma 
at high pressures and its bearing on 

basalt genesis 114 

Pyroxenite stage in basalt genesis . .120 
Isothermal sections of pyroxene quadri- 
lateral 121 

1350°C section 121 

1250°C section 122 

1150°C section 123 

1050°C section 123 

The join diopside-silica 130 

The join diopside-akermanite . . . .132 

Deduction of liquid crystallization paths 
in a five-component oxide system 

containing iron 133 

Upper stability limits of magnesian 

chlorites 136 

The muscovite-chlorite-quartz assemblage 137 
Upper stability of muscovite .... 141 
Low-grade metamorphism of micas in 

pelitic rocks 142 

Dioctahedral micas and related mineral 

groups 142 

Illite-chlorite relations at low tempera- 
tures and pressures 144 

Muscovite-phengite micas . . . .146 

Experimental Petrology at Very High 

Pressures 147 

Petrological constitution of the upper 

mantle 147 

Seismic velocities in the upper mantle . 151 

Melting of pyrolite 152 

Ultramafic rocks 152 

The reaction 4 enstatite + spinel e± for- 
sterite + pyrope 157 

The system enstatite-pyrope . . . .157 

The natural system enstatite-pyrope . 161 

Aluminous enstatites 163 

The system diopside-forsterite-pyrope at 

40 kilobars 165 

High-pressure melting relations for jadeite 

composition 171 

Synthesis and stability of ferrosilite . .174 
High-pressure differential thermal analy- 
sis of a fast reaction with CaC0 3 . 176 

Petrography 179 

I. Discriminant functions and petrographic 

classification 179 

Oceanic-island and circumoceanic ba- 
salts 179 

Classification of lamprophyres ; a pos- 
sible petrographic application of 
multigroup discriminant function 

analysis 182 

Discriminant function coefficients and 

normative calculations . . . .185 
II. Chemical and optical petrography . .186 
Notes on some Mediterranean comendite 

and pantellerite specimens . . .186 
Hyalopantellerite from Cuddia Gadir, 

Pantelleria 186 

Trachyte from near Cantina la Croce, 

Montagna Grande, Pantelleria . 187 
Comendite from hill 142, near Capo 

Sandolo, San Pietro, Sardinia . 189 

Descriptive and genetic significance of 

normative ns 190 

On the relation between suites of 

CIPW and Barth-Niggli norms . 193 

On distinguishing basalt from andesite . 195 

Kersantites and vogesites; a possible 

example of group heteromorphism . 196 

Ore Minerals 199 

The Cu-Fe-S system 200 

The Fe-Pb-S system 202 

Pentlandite 204 

Thermal expansion 204 

Pressure effect on breakdown . . . 206 
Phase relations at low temperatures . . 207 

The Fe-S system 207 

Blaubleibender covellite .... 208 
The Cu-Ni-S system at 200° and 100°C 209 
Studies of Ducktown, Tennessee, ores and 

country rocks 211 

X-ray fluorescence and electron-probe 
analyses of some pyrite-type min- 
erals 214 

Pyrite 214 

Vaesite 214 

Cattierite 214 

Bravoite 214 

Villamaninite 216 

Opaque Minerals in Stony Meteorites . .217 

Sulfide-Silicate Reactions 218 

Anhydrous silicates 218 

Hydrous silicates 221 

Crystallography 222 

Crystal structure of mullite .... 223 
Composition limits of mullite, and the 
sillimanite-mullite solid solution 

problem 227 

Prediction of mica structures from compo- 
sition and cell dimensions . . . 228 
Crystal structures of coexisting muscovite 

and paragonite 232 

Muscovite-paragonite solid solution . 235 
Ferrosilite 237 

Morphological supergroups of structural 

space groups 237 

The ionic radius of lithium .... 238 
Using the cell content to derive the for- 
mula of a mineral from its chemical 
analysis 239 

Geochronology and Isotope Geochemistry . 240 
Isotopic composition of lead in volcanic 

rocks from the Mid-Atlantic Ridge . 241 

Interisland variations 243 

Intraisland variations 243 

Lead isotopes and the age of the earth . 244 
Dating orogenic phases in the Central 

Alps with K-Ar ages on hornblende . 247 
Effect of contact metamorphism on the 

ages of minerals 250 

The microcline-orthoclase transition 

within a contact aureole .... 253 

Biogeochemistry 256 

Hydrocarbons from the low-temperature 

heating of kerogen 256 

The hydrogenation of kerogen from sedi- 
mentary rocks with phosphorus and 
anhydrous hydrogen iodide . . . 258 

Chemicals from the Nonesuch shale of 

Michigan 262 

A geochemical study of some Adirondack 

graphites 265 

Proteins in mollusk shells 267 

Miscellaneous Administration .... 270 

Precambrian Symposium 270 

Journal of Geophysical Research . . .271 

Journal of Petrology 271 

Lectures 271 

Penologists' Club 273 

Summary of Published Work .... 274 

Bibliography 280 

References Cited 281 

Personnel 289 


Earth scientists have long appreciated 
the central role of basalts in the formation 
of the crust. As geophysical data accumu- 
lated it became evident that lavas 
originate deep in the earth (below 60 km) 
and reflect processes occurring there. 
Occasionally eruptions are accompanied 
by peridotite, eclogite, and other nodules 
evidently brought quickly to the surface 
from great depths. Thus, in addition to 
having intrinsic importance in forming 
part of the crust itself, basaltic magmas 
provide a window to the interior of the 

In spite of the importance of these 
rocks, all-out study of them was delayed 
until recently. Fine grain size and 
mineralogical complexity pose formidable 
problems in the separation and identifi- 
cation of phases. 

The presence of ten or more compo- 
nents, including iron with its variable 
valence, complicates the selection and 
control of experimental systems to repre- 
sent the generation and crystallization of 
natural basaltic magmas. 

This year a major fraction of the effort 
of the Laboratory was devoted to basalt 
problems. From an accumulation of 
experimental and analytical data, an- 
swers to the major questions are emerg- 
ing. As a result of work by Schairer and 
Yoder, the basalts can now be delineated 
into significant compositional groups and 
their fractionation can be described in 
terms of a simple flow sheet. 

The flow sheet points up one of the 
most perplexing problems of petrology. 
The major magma types which appear to 
be closely related in their field relations 
are clearly separated by thermal barriers 
at 1 atmosphere of pressure. Various 
schemes have been proposed to bridge 
these barriers, and some new ideas with 
considerable potential are presented this 
year. One idea explored by Bailey in- 
volves the formation of an iron-albite 
molecule, thereby bringing about under- 

saturation in the residual liquids of a 
crystallizing saturated magma. 

Work involving pure components was 
supplemented by Tilley, Yoder, and 
Schairer in studies on melting relations of 
Hawaiian basalts and in experiments on 
analyzed natural pyroxenes. Some of the 
minerals characteristic of the alkali 
basalts including acmite, melilite, wollas- 
tonite, and monticellite were studied by 
Kushiro and Yoder over a considerable 
pressure range. The inverse behavior of 
akermanite and soda melilite, key min- 
erals of the melilite group, with pressure 
is of particular interest because the 
relationship emphasizes the depth control 
of the heteromorphism of a specific rock 
bulk composition. 

A growing amount of geophysical and 
petrological evidence indicates that the 
dominant rock type in the upper mantle 
is an aluminous peridotite. Most of the 
experimental work carried out at high 
pressures this year has been directed at 
increasing our understanding of the 
mineral equilibria in such peridotites and 
of their melting relations. These experi- 
ments show Ringwood, MacGregor, and 
Boyd that a peridotite whose composition 
approaches a mixture of one part basalt 
and three parts dunite can crystallize in 
three facies. These facies, stable under 
different PT conditions, are garnet perid- 
otite, spinel peridotite, and a peridotite 
in which the A1 2 3 is entirely in solution 
in the pyroxene. Probable geothermal 
gradients for oceanic regions and conti- 
nental shield areas indicate that the 
spinel peridotite facies should be present 
beneath the oceans to a depth of about 
60 km but that it should be absent or very 
thin beneath the shields. This speculative 
picture is in harmony with the natural 
distribution of spinel and garnet perido- 
tites that have been erupted from the 
mantle into the crust. Increasing pressure 
causes A1 2 3 in solution in pyroxene to be 
exsolved as garnet. Determination of 




pyroxene-garnet phase relations has pro- 
vided a quantitative explanation of a 
seismic discontinuity in the upper mantle. 
These phase relations also explain the low 
A1 2 3 contents of enstatites from ultra- 
mafic inclusions in kimberlite. Study of 
melting relations in the system diopside- 
forsterite-pyrope (B. Davis) indicates 
that the low-melting fraction in this 
complex "synthetic peridotite" system is 
poor in forsterite and approaches the 
composition of basalt. These observations 
are in accord with results obtained by 
O'Hara (Year Book 62), who employed 
mixtures of analyzed natural minerals. 

Of considerable importance to our 
knowledge of the pyroxenes and basalts 
was the synthesis of the pure ferrous iron 
pyroxene ferrosilite (FeSi0 3 ) by Lindsley, 
MacGregor, and B. Davis. This mineral 
is not stable at high temperatures under 
1 atmosphere of pressure, and is not 
found in nature, though solid solutions of 
(FeMg)Si0 3 are very common. Ferrosilite 
can be readily prepared from fayalite plus 
quartz at temperatures of 1000° to 1300°C 
and 20 to 40 kilobars pressure. Optical 
and crystallographic data for this pure 
end member will serve as an aid to obtain- 
ing determinative curves. Study of the 
stability and melting relations of jadeite 
(Bell) provides further dramatic evidence 
of the changes produced by high litho- 
static pressure in silicate systems. Since 
jadeite has been shown to melt at much 
lower temperatures than diopside it is 
evident that sodium will be concentrated 
in the liquid relative to calcium during 
the melting of rocks in the upper mantle. 
The discovery that the calcite-aragonite 
reaction becomes nonquenchable at tem- 
peratures above a few hundred degrees 
probably explains why aragonite is not 
found in high-grade metamorphic rocks. 

In last year's report it was shown that 
basalts of the circumoceanic and oceanic- 
island environments could be effectively 
distinguished from one another by means 
of Ti0 2 content but that in other respects 
they were so similar that unweighted 
combinations of oxides other than Ti0 2 

would be of little value for this purpose. 
The normative parameters often relied 
upon for intergroup comparisons of this 
type are not unweighted combinations, 
however, and so the earlier work did not 
exclude the possibility of effective classifi- 
cation by more conventional means. 
Chayes now shows that many, and sug- 
gests that all, of the normative param- 
eters now in use are essentially weighted 
linear combinations of oxides. Their 
taxonomic strength may thus be tested by 
discriminant function calculations based 
on the relevant oxides. On the basis of 
rather extensive calculations of this sort 
Chayes and Metais conclude that no 
weighted line or combination of oxides 
that does not include titania is as effective 
as titania alone. The implication is clear; 
consistent and effective discrimination 
between basalts of the two environments 
is possible, but not on the basis of con- 
ventional petrochemical calculations. 

The two-group discriminant function 
having proved effective in the basalt 
study, Chayes and Metais have also been 
experimenting with more complicated 
multigroup discriminant analysis, using a 
large collection of lamprophyre analyses 
as raw material. In the basalt study the 
two classes are defined geographically, 
and the problem is to determine what if 
any relation exists between chemical 
composition and geography. In the 
lamprophyre work the groups are based 
on mineralogy, and it is known from 
inspection that there is much overlap in 
composition. It nevertheless seems pos- 
sible that discriminants based on a few 
key oxides will prove effective. 

In other petrological studies Zies and 
Chayes have continued their work on 
peralkaline silicic lavas and associated 
rocks, and Metais has detected what 
appears to be a rather extreme example of 
heteromorphism between two varieties of 

Gast and Tilton took advantage of an 
opportunity provided by natural isotope 
tracers to investigate the homogeneity of 
rocks derived from the mantle. In 



principle, tracer methods may serve to 
elucidate the processes by which rocks 
themselves originate. The isotopic compo- 
sition of lead and strontium in surface 
rocks is highly variable, depending on the 
rock type and its age. If, on the other 
hand, lead in rocks of deep-seated origin, 
such as basalt, could be shown to have 
narrowly defined isotopic compositions, 
these could serve as tracers to distinguish 
rocks having deep origin from those 
containing large admixtures of surficial 
materials. Volcanic rocks from oceanic 
areas are especially suited to the early 
phases of such an investigation, since the 
possibility of contamination by the radio- 
genic lead and strontium commonly 
associated with a continental crust is 
avoided or minimized. In the past year an 
investigation of the isotopic composition 
of lead in volcanic rocks at Gough and 
Ascension Islands on the mid-Atlantic 
ridge was started. The lead in basaltic 
rocks is not uniform in isotopic compo- 
sition, being much more radiogenic at 
Ascension than at Gough. Moreover, the 
lead from basalt is less radiogenic than 
that from trachyte at each island even 
though both rock types are thought to 
originate in the outer mantle of the earth. 
The data at hand show clearly that it is 
not possible to assign a unique isotopic 
composition to "mantle lead," but rather 
that this information must be established 
for any particular area where work is to 
be done. The basalt and trachyte com- 
parisons indicate the need for caution in 
assuming that rocks are contaminated by 
crustal materials when their lead does not 
have the same isotopic composition as 
lead from basalt. Data for the isotopic 
composition of strontium, obtained in 
parallel with lead, indicate that similar 
arguments apply to strontium. 

In addition to the basalt-related studies 
(only part of which have been mentioned) 
the staff carried out a large number of 

The isotope group has participated in 
several other activities. One investigation 
(Tilton and Doe) dealing with the iso- 

topic composition of lead in feldspars and 
galenas of various ages suggests a value 
for the age of the earth (at least 4.7 X 10 9 
years) different from the currently ac- 
cepted one of 4.55 X 10 9 years. The 
effects of contact metamorphism on 
mineral ages have been studied by 
G. Davis, Tilton, and Hart in order to 
gain better understanding of the response 
of various systems to thermal disturb- 
ance. Detailed information has been 
obtained by examination of rocks from 
near Eldora, Colorado, where a Tertiary 
stock intrudes Precambrian metasedi- 

Another project (Steiger) has involved 
the use of potassium-argon ages from 
hornblende to help decipher the complex 
metamorphic history in the Central Swiss 
Alps. The geochronological results agree 
with petrofabric observations and estab- 
lish the sequence and approximate times 
of phases of the Alpine orogeny. 

Modern scintillation and proportional 
counting techniques, and high-speed digi- 
tal computers, are now the crystal- 
structure analyst's most powerful tools. 
At the Geophysical Laboratory they are 
capable of providing Burnham with X-ray 
diffraction data far superior to those of 
less than a decade ago, and afford him a 
manipulative power capable of examining 
atomic arrangements in the most complex 

Studies on synthetic mullite completed 
this year show it to be a complex defect 
structure with two oxygen sites and two 
cation sites only partially occupied. 
Analysis of the apparent atomic thermal 
vibrations has revealed an oxygen distri- 
bution that varies from unit cell and has 
demonstrated that errors in atomic 
distributions in an assumed structure 
model are easily masked by unrealistic 
thermal parameters. A three-dimensional 
analysis of the structures of coexisting 
muscovite and paragonite by Burnham 
and Radoslovich has demonstrated that 
the difference in the Na/K ratio in the 
interlayer positions of these micas causes 
only minor readjustments of the surface 



oxygen layers and leaves the dioctahedral 
layers essentially unchanged. This study 
also provides a structural explanation for 
the observation that solid solution of 
paragonite in muscovite is far more 
extensive than that of muscovite in 
paragonite. Further progress is reported 
by J. D. H. Donnay and G. Donnay in 
the study of the relation between crystal 
structure and crystal morphology: some 
space groups that cannot be differentiated 
by X-ray diffraction can be distinguished 
by the relative frequencies of the crystal 

A wide liquid immiscibility field has 
been found (Kullerud) to transect the 
entire copper- iron-sulfur system at ele- 
vated temperatures. The new results are 
now being integrated with all previous 
data into a detailed description of the 
complete system from liquidus tempera- 
tures down to 100°C. Continued studies 
of the Fe-Pb-S system (Brett and Kul- 
lerud) at elevated temperatures showed 
that the interruption in the tie lines be- 
tween galena and pyrite at 717° ± 3°C 
is due to the appearance of a liquid field 
which cuts the FeS 2 -PbS join at this 
temperature. This liquid field increases 
rapidly with increasing temperature and 
cuts the PbS-FeS tie lines at 848° ± 3°C. 

Pentlandite of composition Fe4.5Ni4.5Ss 
breaks down at 610°C in the presence of 
vapor. Investigation of this breakdown 
(Bell, England, and Kullerud) under high 
confining pressure with a new differential 
thermal analysis method showed that the 
temperature of decomposition decreases 
with increasing pressure and is as low as 
525°C at 36 kb. These results indicate 
that pentlandite cannot crystallize direct- 
ly from a magma, and they shed entirely 
new light on the occurrence of this 
mineral in ores and meteorites. 

The investigations of mineral assem- 
blages from selected ore deposits have 
continued (Kullerud and Moh). About 
130 sphalerite-pyrrhotite and 40 pyrrho- 
tite-pyrite assemblages were studied from 
various levels of the Calloway Mine, 
Ducktown, Tennessee. The temperatures 

of formation of ore assemblages indicated 
by these methods show systematic vari- 
ations depending not only on depth in the 
mine and position on each mining level 
but also on the associated rock types. 

Kullerud and Yoder have continued 
their interesting work on sulfide-silicate 
systems. At temperatures of 650°C and 
above sulfur reacts relatively rapidly with 
silicates containing iron. Products formed 
depend on the proportions of reactants, 
but in a typical reaction olivine was 
converted into pyrite plus magnetite plus 
enstatite. Studies of the action of sulfur 
on iron-bearing hydrous minerals, con- 
ducted at lower temperatures (500° and 
600°C), yielded some unexpected prod- 
ucts — oxyhornblende and oxymica, phases 
that have been observed in ore deposit 

Layer silicates (micas, chlorites, clays) 
play a most important part in the reac- 
tions of diagenesis and low-grade meta- 
morphism. The wide compositional vari- 
ations possible in these mineral groups 
render them difficult to study experi- 
mentally. Fawcett and Velde have deter- 
mined the upper stabilities of the 
important end members Mg chlorite and 
muscovite. Velde has determined pressure 
and temperature stabilities of solid solu- 
tions with muscovite which approximate 
natural illites and phengites, and Faw- 
cett is examining possible reactions in 
the assemblage muscovite-chlorite-quartz. 
Their data are beginning to show the 
importance of layer-silicate solid solution 
relationships in reactions producing bio- 
tit e and garnet. 

A major problem in phase equilibria 
studies of systems of four or more compo- 
nents has been the presentation of data 
in useful graphical form. Presnall has 
shown that it is possible, by means of 
phase diagrams, to deduce liquid crystal- 
lization paths in a five-component system 
containing FeO and Fe 2 3 . This method 
of presentation should be of importance 
in the study of systems approximating 
some mafic and ultramafic rocks. 

Several new developments arose from 



biogeochemical research. Hoering and 
Abelson found saturated fatty acids in the 
billion-year-old Nonesuch shale at White 
Pine, Michigan. They also obtained 
saturated hydrocarbons from a Precam- 
brian petroleum at this locality and 
extracted a similar suite of hydrocarbons 
from rocks of the Stripy Marker member. 
Hoering investigated the hydrogenation 
of kerogen by phosphorus plus anhydrous 
hydrogen iodide. By this means he was 
able to obtain abundant quantities of 
hydrocarbons from Precambrian rocks 
previously considered barren. 

Hare and Abelson have examined the 
proteins associated with molluskan shells. 
Next to the arthropods the mollusks are 
represented by more species than any 
other phylum, yet except for a few 
advanced forms all have a shell consisting 
principally of calcium carbonate in the 
form of calcite, aragonite, or a combina- 
tion of the two. The small mineral crystals 
are bonded by layers of protein. With 
this combination of materials the mol- 
lusks, including gastropods (e.g., snails), 
pelecypods (e.g., clams), and cephalopods 
(e.g., pearly nautili), are able to produce 
tens of thousands of different sizes and 
shapes. It seemed possible that the pro- 
tein bonding material might play a key 
role and that an investigation of their 
amino acid content might yield significant 

Since the shape of the shell is geneti- 
cally controlled, we can guess that the 
protein bonding material also might be 
under genetic control. The amino acid 
content of about one hundred different 
shells has been studied, and the proteins 
of related species have been found to have 
similar composition. Large differences 
have been noted in species not closely 
related. Analysis of amino acids in 
molluskan shells provides a new tool for 
phylogenetic investigations. 

One of the principal scientific argu- 
ments advanced for exploration of the 
moon is the expectation that much will be 

learned about the origin and history of 
the solar system. Study of meteorites 
promises to provide information at least 
equally relevant and in some ways more 

Samples of extraterrestrial origin, which 
frequently fall on earth, provide samples 
from vast volumes of space, but study of 
these objects has been sporadic. At the 
moment there is increased activity, 
especially in study of the effects of cosmic 
rays on the meteorites and in chemical 
investigations. In the past there have 
been some mineralogical studies, particu- 
larly on the iron and stony-iron meteor- 
ites. There has been little systematic 
investigation of the opaque minerals in 
stony meteorites. Recent advances in 
techniques, especially in preparation of 
polished sections, have made these opaque 
minerals accessible to expert optical 
examination by Ramdohr. With the new 
preparations it was possible for him to 
study fine textures and to identify 
minerals present in only trace amounts. 
As a result numerous minerals known 
from the earth have been observed for the 
first time in meteorites. In addition, 
during the past three years about twenty 
entirely new minerals have been dis- 
covered. They include bizarre sulfides and 
arsenides, for example (Mg,Mn,Fe)S. 
Since 1960-1961 a total of 340 polished 
sections of 240 different falls and finds 
have been investigated and described. All 
these observations have now been com- 
piled into a manuscript which together 
with nearly 300 photomicrographs is 
scheduled for publication by the Smith- 
sonian Institution. All polished sections 
from which the photographs and descrip- 
tions were made have been placed in 
custody of the Smithsonian Institution, 
where they will be available to qualified 

Discussion of some of Ramdohr's 
observations as well as detailed descrip- 
tion of other work of the Laboratory 




An Evaluation of the New Progress 

Only a few of the plethora of ideas that 
evolve from experiment and study come 
to fruition within the annual report year, 
yet those recorded here represent the 
culmination of many years of intensive 
effort on several mainstream problems of 
petrology. The basalts, long considered 
the key to the evolution of most igneous 
rocks, can now be delineated into signifi- 
cant composition groups, and their frac- 
tionation can be described in terms of a 
simple flow sheet. Never before has it been 
possible to outline on the basis of experi- 
mental results the overall plan that 
controls the fractionation of the major 
magma types. The construction of this 
plan resulted from the determination of 
the phase equilibria of a vast array of 
compositions specific to alkali basalts. 
The vital observation drawn was that the 
many important rock types appear to 
express the univariant and invariant con- 
ditions of nature, and are not just random 
accumulations of minerals. Some of the 
minerals characteristic of the alkali 
basalts and their derivatives required 
detailed study. Among them are acmite, 
melilite, wollastonite, and monticellite, 
for which new data, involving a consider- 
able pressure range, are presented. The 
inverse behavior of akermanite and soda 
melilite, key minerals of the melilite 
group, with pressure is of particular 
interest, because the relationship empha- 
sizes the depth control on the hetero- 
morphism of a specific rock bulk compo- 

The flow sheet points up one of the 
most perplexing problems of petrology. 
The major magma types, which appear to 
be closely related in their field relations, 
are clearly separated by thermal barriers 
at 1 atm. Various schemes have been 
proposed to bridge these barriers, and 
some new ideas with considerable poten- 
tial are presented this year. One idea 
involves the formation of an iron-albite 

molecule, thereby bringing about under- 
saturation in the residual liquids of a 
crystallizing saturated magma. A very 
powerful concept involves the separation 
of a pyroxene from the magma at the 
initial stages of crystallization. The 
pyroxene must, of course, be the normal 
pyroxene crystallizing at the depth en- 
visaged and must be effective over a 
temperature range in which that pyroxene 
normally crystallizes as the dominant 
phase. The nature of the pyroxene that 
can be extracted and give rise to a crit- 
ically undersaturated residual liquid from 
an undersaturated magma has been 

The concept involving the extraction of 
a specific pyroxene becomes most impor- 
tant in the light of the confirmation that 
basalt in transforming to eclogite with 
increasing pressures passes through a 
pyroxene-rich stage. Experiments on the 
pertinent mineral joins clearly illustrate 
that the various basalts transform to 
eclogites at considerably different con- 
ditions via unique reactions involving 
primarily pyroxenes. 

The pyroxenes stable at great depths 
are quite different from those in the rocks 
formed at or near the surface of the earth. 
The efforts of the past years in unraveling 
pyroxene complexities at 1 atm are still 
under way. Because of their immense 
importance as a mineral group the details 
of their interrelationship are being exam- 
ined at significant temperature levels. 
The initial efforts at high pressure for one 
of the important pyroxene joins are also 
presented and should be related to the 
work on jadeite outlined in a following 
section of the report. 

The two most important minerals in 
the initial stages of metamorphism are 
chlorite and muscovite. Identification of 
their precursors and the nature of the 
reactions that eventually lead to their 
consumption lies at the heart of the prob- 
lems of metamorphic rock formation. The 
sluggish behavior of the materials makes 



progress slow; yet the results will form 
the foundation for the new concepts in 
progressive metamorphism. 

Crystal and Liquid Trends in 
Simplified Alkali Basalts 
J. F. Schairer and H. S. Yoder, Jr. 

Some years ago Yoder and Tilley 
(unpublished work in 1958) constructed a 
simple iron-free tetrahedron based on the 
principal normative minerals of various 
basalts. An extensive experimental pro- 
gram was initiated to provide funda- 

mental information on the faces of this 
tetrahedron and two of the principal 
internal planes (Fo-Di-Ab and Ab-Di-En) . 
This tetrahedron appeared first in Year 
Book 59 (p. 67, fig. 15) and was general- 
ized to include members commonly in 
solid solution (p. 68, fig. 16), and data 
were given there on two of the faces of 
the tetrahedron and one internal plane. 
The origin of basalt magmas was dis- 
cussed by Yoder and Tilley in a major 
publication (Yoder and Tilley, 1962) with 
the aid of this basalt tetrahedron. 

Mo I per cent 

Fig. 1. Expanded basalt tetrahedron nepheline (Ne)-forsterite (Fo)-silica (Si0 2 )-larnite (La). 
Dotted planes enstatite (En)-albite (Ab)-diopside (Di) and albite-diopside-wollastonite (Wo) are 
planes of silica saturation. The planes Di-Wo-Ne and Di-Fo-Ne separate albite from akermanite 
(Ak). Mo, monticellite ; Mer, merwinite; Ra, rankinite; Sm, soda melilite; Qz, quartz. 



When experimental data had been 
obtained on the faces and principal limit- 
ing internal planes it became obvious that 
this tetrahedron had to be expanded to 
follow the courses of crystallization in 
alkali basalts. The expanded tetrahedron, 
Ne-Fo-Si0 2 -Ca 2 Si04 (La), is given here as 
figure 1. The previous simplified basalt 
tetrahedron becomes a smaller tetra- 
hedron within this expanded tetrahedron. 
Even the expanded tetrahedron is not 
adequate to account for all solid solutions 

(e.g., anorthite molecule in plagioclase). 

During the past year a major experi- 
mental attack has been made on a sub- 
stantial portion of this expanded tetra- 
hedron. We present here some of the 
experimental results. 

Figure 2 gives the phase-equilibrium 
data for the join Ne-Ak-Di. Two piercing 
points, 01 + Mel + Di ss + L (1213° ± 
5°C) and Ne ss + Mel + 01 + L (1178° 
± 5°C) appear in this join. 

Figure 3 gives the phase-equilibrium 

1454 + 2' 

Ca 2 MgSi 2 7 



Na Al Si0 4 

1343° Ca Mg Si 2 6 


Weight per cent 

Fig. 2. Phase-equilibrium data for the join Ne-Ak-Di. The lower-case letters indicate the tem- 
perature of appearance of a phase with decreasing temperature for a specific bulk composition. The 
first number is the liquidus temperature, and the phase is indicated by the field in which it lies, 
m, melilite; d, diopside; f, olivine; n, nepheline; eg, carnegieite; and BM, beginning of melting. Addi- 
tion of the letter o indicates that the phase has been consumed and is no longer present. All 
phases are solid solutions. 




Ca 2 MgSi 2 7 





I70±5° 1218' 
, 236 . BMII70i5->7 
ego 1222' 
8M Il70i5' 



Na AlSiO. 

Na CaAISi 2 7 

J544 + 2° 
Ca Si 0, 

Weight per cent 

Fig. 3. Phase-equilibria data for the join Ne-Ak-Wo. Abbreviations as in figure 2, as well as w, 
wollastonite; p, pseudowollastonite. Soda melilite composition is Ne = 55.02 and Wo = 44.98 in 
weight per cent. See figure 20 for details of akermanite-soda melilite join. 

data for the join Ne-Ak-CaSi0 3 . The 
melilite join akermanite (Ca 2 MgSi 2 7 )- 
soda melilite (NaCaAlSi 2 7 ) appears in 
this join, and there is one piercing point, 
Ne ss + Mel + Pwo + L (1171° ± 3°C). 
Cf. figure 20. 

Figure 4 gives the phase-equilibrium 
data for the join Ab-CaSi0 3 -Ak. Three 
piercing points appear in this join: 
Mel + Di S3 + Wo + L (1268° ± 3°C), 
Mel + Wo + Pwo + L (1320° ± 5°C), 
and Plag + Wo + Di ss + L (1095° ± 

Figure 5 gives the phase-equilibrium 
data for the join Ab-Ak-Di. No piercing 
points lie in this plane. 

Figure 6 gives the phase-equilibrium 

data for the join Ne-Ak-Ab. Two piercing 
points, Ne S3 + Mel + Di s3 + L (1100° 
=fc 5°C) and Plag + Ne ss + Di S3 + L 
(1083° d= 5°C), lie in this join, and atten- 
tion is called to the proximity in compo- 
sition of these two points, each of which 
must lie close in composition to a quater- 
nary invariant point which must also lie 
a short distance within the volume 
Ne-Ab-Ak-CaSi0 3 . The location of these 
points within this volume will be dis- 
cussed in more detail later in this report. 
Figure 7 is a preliminary phase- 
equilibrium diagram for the join Ne-Wo- 
Di, showing the rough positions of the 
several piercing points. Additional data 
to delineate the precise compositions and 


Ca Si 0, 



Na Al SLO. 

3 8 d 1101° 


-pi 1 127' 


Ca Mg Si, 0, 

Weight per cent 

Fig. 4. Phase-equilibria data for the join Ab-Wo-Ak. 
well as pi, plagioclase. 

Abbreviations as in figures 2 and 3, as 

temperatures of the several piercing 
points are now being obtained. 

Attention is called to the following 
previously recorded data: (1) Ne-Fo-Si0 2 
(Schairer and Yoder, Year Book 60, p. 
142, fig. 35); (2) Fo-Di-Si0 2 (Kushiro and 
Schairer, Year Book 62, p. 100, fig. 25); 
(3) Ne-Fo-Di (Schairer and Yoder, Year 
Book 59, p. 70, fig. 18); (4) Ne-Di-Si0 2 
(Schairer and Yoder, 1960) ; (5) Fo-Ab-Di 
(Schairer and Morimoto, Year Book 57, 
p. 213, fig. 24); (6) Ab-En-Di (Schairer 
and Morimoto, Year Book 58, p. 115, fig. 
16); (7) CaSi0 3 -Ak-Di, CaSi0 3 -Di-Si0 2 , 
Ak-Fo-Di (Ferguson and Merwin, 1919); 
(8) Ne-Ak-CaSi0 3 (Foster, 1942). 

These previously recorded data and the 

new data on six joins presented here in 
figures 2 through 7 permit the construc- 
tion of a flow sheet (Schairer, 1942, 1954) 
showing the relations between the quater- 
nary invariant points in the geologically 
important portion of the expanded basalt 
tetrahedron. This flow sheet is given here 
as figure 8. All the solid phases except 
pure Si0 2 itself and possibly the high- 
temperature form of CaSi0 3 , pseudo- 
wollastonite, are complex solid solutions. 
Considerably more experimental data are 
needed to adequately define the precise 
composition of these solid solutions, all of 
which are rock-forming minerals. Because 
of these complex solid solutions many 
compositions crystallize completely to 



1454 + 2° AKERMANITE 

: 90 



M E L I L i T E 

I423 a 



m I273°_ 

w 1238° 

BM 1073+5° 

c/-),/ 1263° 

0U /T w 1198° 

/ pi 1088° 

/ BM 95515° 

m I3I3V 


BM 1133+5° 

A 50 
\ ^1361.5 

70 A 

DU /\ „ ||53° 

/ pi 1103° 
/ BM 94815° 

pi 1078° 
BM 990+5° 



1205° / 
pi 1113' / 

w 1098° / 
BM 950 + 5°/ 



1 P s 

1 D 





1138° /pi II 28° 
d 1127°. / 
BM995+IO°\ „„ Z< if. 

,| 4| . \90^fP" 138 ' * 

BM990 + I0' ^v/ \BM 1033+5° 

BM 983 + 5° 

__- — pi 1128" 

< BM993±5° 




/ V 

I V 




\ 1388° 

Y \l39l.5 

PLAGIOCLASE— p!^ \«^-U40 

1 1 18+3°/ ,\/ p '" 33 

• BM 1033+5° 

ALBITE \ f MO^i 
NaAISLO. ) Si4B»\ l Ji l 

Il43°/ dl096 * dl!23° 
d 1068° 

58° 20 


40 50 

Weight per 





1368° CaMgSi 2 6 

Fig. 5. Phase-equilibria data for the join Ab-Ak-Di. Abbreviations as in figures 2, 3, and 4. 

only three solid phases instead of pro- 
ceeding to a quaternary invariant point 
with four solid phases. 

The temperature maxima in AF and 
GE (fig. 8) separate the quaternary 
invariant points A, B, and G in the 
siliceous portion of the expanded basalt 
tetrahedron (fig. 1) from the quaternary 
invariant points F, C, D, and E which lie 
in the nepheline-rich portion of this 
tetrahedron. The temperature maximum 
in BG separates the quaternary invariant 
point G from the quaternary invariant 
points A and B. The quaternary invariant 
point B with plagioclase, clinopyroxene, 
orthopyroxene, and tridymite is the 
principal goal in geologically important 
compositions in the siliceous portion of 

the expanded basalt tetrahedron. 

The quaternary invariant point E with 
sodic plagioclase, clinopyroxene, nephe- 
line, and wollastonite solid solution is the 
principal goal in silica-poor (nepheline 
normative) portions of the expanded 
basalt tetrahedron, but many melts may 
crystallize completely before they reach 
this goal. Each of the quaternary invari- 
ant points represents an important rock 
type achieved by crystallizing melts 
especially when fractional crystallization 
is operative. A temperature maximum in 
FC (fig. 8) provides two paths toward the 
quaternary invariant point E, one from 
F to E and the other from C through D 
to E. 

On the basis of the observations at 1 




n \l1H BM \'\7Q±S 

ego 1222° TTfJL 

208° A^\ 


Na Al Si 4 


Na Al 

Weight per cent 
Fig. 6. Phase-equilibria data for the join Ne-Ab-Ak. Abbreviations as in figures 2, 3, and 4. 

atm described above, some general appli- 
cations to natural rock systems may be 
made. Of course, the laboratory observa- 
tions are restricted to a portion of the 
Na 2 0-CaO-MgO-Al 2 3 -Si0 2 system, and 
it is known that additional A1 2 3 as well 
as K 2 0, FeO, Fe 2 3 , Ti0 2 , and a host of 
rare elements play their roles in the 
evolution of the alkaline rocks. In spite of 
compositional limitations, key problems 
in alkaline rock genesis may be delineated 
more distinctly. 

The magma type recognized by most 
field workers as the parent of the alkali 
succession is alkali basalt magma (Yoder 
and Tilley, 1962, p. 346). It is composed 
predominantly of clinopyroxene, plagio- 
clase, and olivine when crystalline, and 

may contain small amounts of occult 
nepheline or hypersthene. l Assuming that 
such a parental magma can be generated 
on or near Di-Fo-Ab, for example by the 
schemes outlined by Yoder and Tilley 
(1962) or Tilley, Yoder, and Schairer 
(this report), the first derivative rock 
according to the flow sheet (fig. 8) is most 
likely to be a nepheline basanite as 
represented by invariant point F. The 
rock names relating to the mineral assem- 

1 As the result of the careful work of Kushiro 
and Schairer (Year Book 62, pp. 95 ff.) the 
equilibrium thermal divide now lies on a plane 
slightly inclined to Di-Fo-Ab involving a com- 
plex solid solution of Di in which En as well as 
Ca-Tschermak's molecule and Fo may be 
important, monticellite in Fo, and anorthite in 



Ca Si 0, 





I490 c 

NaCaAISi 2 7||98 

1164+25/ __^ 


NaAISi0 4 

1258+3° 1302° 

1343° CaMgSi 2 6 

Weight per cent 

Fig. 7. Preliminary phase-equilibria data for the join Ne-Wo-Di. Abbreviations as in figures 2, 
3, and 4. 

blages are presented in a similar flow 
sheet (fig. 9), corresponding to the 
univariant lines and invariant points in 
figure 8. Basanites are among the more 
important alkaline rocks. As a result of 
reaction, olivine is consumed 2 and a 
magma giving rise to nepheline tephrites 
is probably produced. Such a rock type 
would represent liquids between the 
points F and E on the flow sheet (fig. 8). 
With further fractionation some liquids 
may reach eutectic point E (fig. 8), and 
rocks having that rare assemblage of 

2 The existence of an olivine reaction relation- 
ship in alkali rocks was first pointed out by 
Schairer and Yoder (1960) and as yet is not 
widely recognized. 

minerals would be called wollastonite- 
nepheline tephrite. More likely the wol- 
lastonite component would be represented 
by perovskite or sphene in natural rocks. 
As is noted in the flow sheet (fig. 8) the 
eutectic point E is also reached by a 
completely independent, but less impor- 
tant, course of fractionation. Such a 
course of fractionation is recorded by a 
rare group of rocks. The parental liquid 
of the second course of fractionation may 
be represented either by an olivine 
nephelinite which lies on or near the 
Di-Fo-Ne plane or by an olivine melil- 
itite, the so-called melilite basalt, which 
lies on or near the Di-Fo-Ak plane. These 
liquids descend to point C, where the 




(Ab-Fo-Si0 2 ) 
1098 ±10° 

(Fo-Di-Si0 2 ) D|ss 
I386±l° p' R SS 


/ Diss 









~ OL 














(Ab-Fo-Si0 2 ) 


(Fo-Di-Si0 2 ) 





Temp ? 





( Wo-Ab-Si02) 






(Wo-Di-Si0 2 ) 


£ 950+5° 







(Ne- Ak -Di) 
-1213 + 5° 




NEss ni 

Cil40±5- 0L 



1178 + 5° 

NEss 'h 






/ W0 DI 
Dl065±5° M £ 


wo hr 

(Wo- Ak-Di) 


1171 + 5° 



Ne- Ak-Wo) 

Fig. 8. Flow sheet of liquids in a major portion of the expanded basalt tetrahedron Ne-Fo-Si0 2 -La 
based on six new planes (figs. 2, 3, 4, 5, 6, 7) and published data listed in text. OL, olivine; DI, 
diopside; PR, protoenstatite; PL, plagioclase; WO, wollastonite; TR, tridymite; NE, nepheline; 
MEL, melilite. Capital letters indicate invariant points; lower-case letters are for convenience in 

assemblage 01 + Mel + Ne ss + Di ss , an 
olivine-melilite nephelinite, is the repre- 
sentative rock type. Another olivine 
reaction takes place with liquid, and 
when it is consumed the liquids, if 
crystallized, would be represented by 
melilite nephelinites. Such liquids when 
joined by wollastonite at point D (fig. 8), 
although exceptionally rare, are called 
wollastonite-melilite nephelinites if crys- 
tallized. As has already been mentioned 
the wollastonite may be taken up by 
perovskite or sphene. More likely the 
wollastonite might combine with some 

nepheline to form the soda-melilite mole- 
cule, CaNaAlSi 2 7 (cf. soda-melilite sec- 
tion, this report). At this point in the 
fractionation sequence an important re- 
action takes place, the reaction of melilite 
with liquid. 3 On consumption of melilite, 
the liquid, which may carry the phases 
wollastonite, nepheline, and clinopy- 
roxene, is free to move toward point E 

3 Unfortunately, there is no petrographic evi- 
dence to support this reaction relationship in 
natural rocks, and, in general, the field observa- 
tions do not give a clear picture of the relation 
of melilite-bearing to plagioclase-bearing rocks. 







Basanite Olivine 



Wollastonite -nepheline 
Tephrite Wollastonite 



Olivine — melilite 



Wollastonite -melilite 

Fig. 9. Rock nomenclature diagram corresponding to flow sheet in figure 8. For the most part, 
names apply to extrusive rock types. Invariant point G may be represented in nature by some 
metamorphic rocks. 

(fig. 8). Rocks having that assemblage, 
called wollastonite nephelinites, are very 
rare. The significant observation is that 
such a liquid on reaching E would pre- 
cipitate a sodic plagioclase. That is, a 
liquid that initially gave rise to a melilite 
could through reaction and fractionation 
yield a plagioclase. 

The clear distinction between the two 
courses of fractionation of the alkali 
parental magmas represented in the 
limited system portrayed here is that one 
liquid gives rise to plagioclase and the 
other to melilite over most of its course. 
It is evident in the natural igneous rocks 
as well as in the laboratory experiments 
that akermanitic melilite and plagioclase 
are incompatible. 4 The incompatibility 

4 Rankin and Wright (1915) demonstrated that 
gehlenite and anorthite are stable together and 
they are indeed found in some contact aureoles. 
The melilites of the igneous rocks lie mainly on 
the akermanite-soda melilite join (see fig. 18). 
It would be of interest to establish the limits of 
solid solution for association of melilite with 
plagioclase from both the chemical analysis of 
natural melilites in equilibrium with plagioclase 
and the synthesis of various combinations of 
Ak-Geh-Sm and Ab-An. 

arises on or near the plane Ne-Fo-Di, all 
members of which are solid solutions, 
separating akermanite and albite. Be- 
cause of this relationship the olivine 
nephelinites are of critical interest in 
determining whether or not the melilite 
succession and the plagioclase succession 
have a common parentage at depth 
(higher pressures). Spinel may play a role 
in this connection. The plane Ne-Wo-Di 
(fig. 7) also separates akermanite and 
albite; however, it has been demonstrated 
that derivative liquids may penetrate this 
plane through the reaction relation of 
melilite with liquid. The phase-equilibria 
data at 1 atm indicate that these two 
successions must arise independently, at 
least in the lavas. 

The alkali successions when combined 
with the successions of the olivine 
tholeiites and quartz tholeiites (expressed 
by invariant points A and B, fig. 8) 
illustrate a vast array of common and 
important rock types. The thesis is 
advanced here that the major rock types 
lie on or near the critical univariant 
curves and the principal invariant points 
of the natural multicomponent system. 



Documentation of this concept will, of 
course, require a large amount of modal 
and chemical data on natural rocks. 
However, it may be assumed that the 
physicochemical behavior of the olivine 
tholeiites outlined by Tilley, Yoder, and 
Schairer (this report) is an example of the 
essential univariant nature of most 
derivative magmas. 

The Peralkaline Residua System: 
Na 2 0-Al 2 03-Fe 2 03-Si0 2 

D. K. Bailey and J. F. Schairer 

During the past year the study of the 
system Na 2 0-Al 2 3 -Fe 2 3 -Si0 2 at atmos- 
pheric pressure was brought to comple- 
tion, and the crystallization flow diagram, 
summarizing the invariant and univariant 

(AIjjOj -Fe 2 3 - SiO;, ) 

Fig. 10. Planar diagram showing the univariant and invariant equilibria involving liquid in 
Na 2 0-Al 2 03-Fe 2 3 -Si0 2 . Univariant lines, with three solid phases plus liquid, link the ternary 
invariant points of the bounding ternary systems (triangles) and the quaternary invariant points 
(octagons). Arrows indicate falling temperature. Only the phase relations may be read from this 
diagram; the geometric arrangement is arbitrary: the relative positions of the piercing points 
(numbered) and the joins (thin broken lines) are shown for convenience. Solid phases: ab, albite; 
ac, acmite; eg, carnegieite; cor, corundum; cris, cristobalite; ds, sodium disilicate; hem, hematite; 
mul, mullite; ne, nepheline; ns, sodium metasilicate; qu, quartz ; trid, tridymite; 5-1-8, 5Na 2 0-Fe 2 3 - 
8Si0 2 . 



equilibria, is shown in figure 10. The full 
description of this study is now ready 
(Bailey and Schairer, 19646), and the 
chief features of interest will be briefly 
outlined here. This system may be 
regarded as the peralkaline extension of 
Petrogeny's Residua system (Bowen, 
1937; Schairer, 1950). It provides data on 
liquids akin to natural peralkaline neph- 
eline syenites and granites (A and B, 
respectively, in fig. 10), and it also offers 
new insight into the nature and relation- 
ships of the rock that has a key role in the 
major continental alkaline complexes — 
namely, ijolite. In addition, a process has 
been found in this system whereby the 
transition from oversaturated to under- 
saturated liquids could be accomplished, 
by substitution of iron for alumina in the 

Ijolite and Nephelinite as Residual Liquids 

One of the most interesting problems of 
the alkaline rocks is the origin of the 
nepheline-pyroxene rocks so character- 
istic of the major continental alkaline 
complexes. In last year's report (Year 
Book 62, fig. 48) we indicated that the 
liquid at the quaternary reaction point 
acmite + hematite + nepheline + albite 
+ liquid (E in fig. 10) was close to the 
nepheline-acmite join and hence analo- 
gous to ijolite. There is clear evidence 
that the ijolite in many alkaline com- 
plexes was mobile and intrusive, and in 
many nephelinite lavas acmite and nephe- 
line were the main phases crystallizing 
during the closing stages of solidification. 
Thus the experimental results, and the 
rocks themselves, leave no doubt of the 
existence of ijolitic liquid, and the 
nephelinite petrography suggests that 
this is a residual liquid, in agreement with 
the finding of an invariant liquid of ijolite 
type in the synthetic system. Because of 
this residual nature, ijolitic liquids might 
be expected to be derived from a range 
of bulk compositions in the same way that 
granitic and nepheline syenitic liquids 
represent the residua of other composition 
ranges. It is of particular interest that all 

three types of residua are found in the 
Na 2 0-Al 2 3 -Fe 2 3 -Si0 2 system, making 
it the peralkaline residua system, wherein 
may be seen the interrelationships of 
ijolite (nephelinite), peralkaline nepheline 
syenite (phonolite), and peralkaline gran- 
ite (pantellerite, comendite). 

Attention was drawn in the last report 
to the close proximity, both in tempera- 
ture and in composition, of the "ijolite 
point" and the ternary eutectic nepheline 
+ acmite + hematite + liquid, desig- 
nated E and (15), respectively, in figure 
10. This proximity means that liquids 
near E, by minor fluctuations in condi- 
tions, could be set on a trend of iron 
enrichment and increasing basicity toward 
eutectic F, i.e. a "melteigite" trend. This 
possibility and others derivable from the 
quaternary relations are summarized in 
the schematic diagram, figure 11. 

The chief purpose of figure 11 is to 
indicate the liquid paths leading from the 
"ijolite point," and it is clearly inade- 
quate for showing all the courses by which 
ijolite itself might be derived. A con- 
sideration of the genesis of ijolite will 
appear with the systematic treatment 
of Na 2 0-Al 2 3 -Fe 2 3 -Si0 2 (Bailey and 
Schairer, 19646), a summary of which 
must suffice here. Ijolite is a common, 
almost invariable, associate of carbonatite 
(W. C. Smith, 1956) and in some com- 
plexes appears to be formed by melting 
of metasomatized country rocks (von 
Eckermann, 1948). Such observations 
have led to the growth of the concept that 
all ijolite is rheomorphic fenite. It was to 
be hoped that the granite controversy 
would have cautioned against such an 
extreme view. The analogy with granite 
may be more apposite than it seems at 
first glance, because if ijolite is a residual 
liquid it will form by partial melting from 
a range of bulk compositions, and being 
rich in volatiles it will presumably tend 
to transform wall rocks to like composi- 
tions, in a similar manner to the process 
of contact granitization. Generation of 
ijolite by partial melting of nepheline- and 
acmite-bearing fenite could be predicted 



From Ab-Ac-Hem Plane 
(possible derivation from 
iron-rich syenite) 

To Quaternary 

Eutectic F 

(melteigite trend) 

From Ab-Ne-Hem Plane 
(possible derivation from 

To Quaternary Eutectic 
(malignite — *- nepheline 
syenite trend) 

Fig. 11. Possible petrologic significance of the quaternary point E- 
four univariant curves that stem from it. 

-the "ijolite" point — with the 

from the existence of an "ijolite" invari- 
ant point in Na 2 0-Al203-Fe 2 03-Si02, but 
it is precisely because this composition is 
a residuum that other sources of ijolite 
must also be expected. The nephelinites, 
for instance, are commonly associated 
with the strongly undersaturated melilite 
basalts, and therefore ijolite would appear 
to be the natural liquid residuum of this 
family of rocks. Moreover, such a deriva- 
tion is consistent with its appearance in 
the carbonatite association, because this 
is linked, through alnoite, to kimberlite, 
which appears to be the deep-seated 
equivalent of melilite basalt. Typical 
kimberlites, too, show transitional charac- 
teristics toward carbonatite (Daly, 1925; 
Dawson, 1962). 

It is possible from these relationships 
to construct a coherent picture of the 
igneous activity involving undersaturated 
alkaline rocks, based on the concept of 
ijolite and nepheline syenite being silicate 
residua, with carbonatite as a volatile- 
rich fugitive fraction from the mantle. A 
schematic presentation of these relation- 
ships is shown in figure 12. 

The localities used as examples in figure 
12 were chosen from Africa, where there 

is a distinct correlation of this type of 
activity with the rift pattern (Bailey, 
1964a). The rifts are located along the 
axes of long crustal upwarps, and it is 
suggested (Bailey, 1964a) that generation 
of magmas of residual type and the 
collection of the fugitive constituents in 
the underlying mantle result from relief 
of lithostatic load below the rift arches. 
This concept provides a structural frame- 
work for the relationships shown in 
figure 12, emphasizing the residual nature 
of the ijolites, syenites, and carbonatites, 
and indicating a possible reason why this 
type of igneous activity should charac- 
terize the stable continental areas of the 
earth's crust. 

Transitions from Over saturated 
to Undersaturated Compositions 

In Year Book 62 we indicated that 
fractionation of early-formed hematite 
from a limited range of undersaturated 
compositions would yield oversaturated 
residual liquids. This extends the origi- 
nal observation of this possibility in 
the system Na 2 Si03-Fe 2 03-Si02 (Bowen, 
Schairer, and Willems, 1930) and sup- 
ports the suggestion of Tilley (1958) that 


Northern Rhodesia Basutoland Eastern Uganda Western Rift Eastern Rift 






Case IB 

MB-* I+IC) 

Case 3 





Case 4 

Fig. 12. The common surface relationships between carbonatite (C), ijolite (I), nepheline syenite 
(phonolite, P) are shown in simplified form and are interpreted in terms of C, I, and P as residua; 
C is the volatile-rich, low-temperature residuum from the mantle; I is the low-melting fraction 
from highly undersaturated rocks such as melilite basalt (MB); and P is the low-melting fraction 
from alkali basalt. The cluster of arrows at the base of each column represents collection of fugitive 
constituents from the underlying and surrounding mantle. Case IB represents transport of kimber- 
lite (K) from the mantle with no melting under crustal conditions. All the examples are African to 
give a unified picture, but they could be matched elsewhere. 

the incongruent melting of acmite could 
effect the transition from undersaturated 
to oversaturated liquids. 

The reverse transition, from over- 
saturated to undersaturated liquids, has 
always been problematical, both experi- 
mentally and from the standpoint of 
available natural examples, and it is 
indeed difficult to envisage if the initial 
liquid were strongly oversaturated. A 
possible mechanism applicable to a 
slightly oversaturated parent has, how- 
ever, become evident from the present 
study; it is treated more fully in the 
systematic discussion of Na20-Al 2 3 - 
Fe 2 3 -Si0 2 (Bailey and Schairer, 19646), 
but as it concerns a fundamental issue in 
petrology it can be outlined here. As was 
indicated last year, the albite crystallizing 
in this system is a solid solution, largely 
of the "iron albite" molecule (NaFe 3+ - 
Si 3 8 ) in albite. This structure requires 
more silica than would be available from 
saturated liquids containing potential 
acmite + albite, and it follows that early 
separation of iron-bearing albite from 

slightly oversaturated liquids could pro- 
duce residual liquids that would cease 
crystallization only at the undersaturated 
eutectic A, albite + acmite + nepheline 
+ sodium disilicate + liquid. In natural 
liquids of quartz syenite composition such 
a process could be still more effective, 
because in potassic feldspars iron can 
substitute for alumina more readily, as is 
demonstrated both by the natural potash 
feldspars (Deer, Howie, and Zussman, 
1963) and by the fact that the pure iron 
analogues of orthoclase (Faust, 1936), 
microcline, and sanidine (Wones and 
Appleman, 1963) can be synthesized, 
whereas the stable assemblage for pure 
"iron albite" is acmite + quartz. The 
possibility of a quartz syenite liquid 
yielding nepheline syenite by this process 
merits attention because it involves 
crystallization of the dominant mineral 
component in the parent, and therefore 
only minor amounts of iron substitution 
during the main phase of crystallization 
would be sufficient to yield a nepheline 
syenite residual. 



Peralkaline Liquids and the Norm 

The two eutectics A and B, figure 10, 
correspond to simplified peralkaline neph- 
eline syenite (phonolite) and peralkaline 
granite (pantellerite) and are noteworthy 
not only for their low temperatures but 
also for the liquid compositions. Both are 
rich in sodium silicate and in consequence 
are well removed from Petrogeny's Re- 
sidua system, KAlSi0 4 -NaAlSi0 4 -Si0 2 , as 
we indicated in Year Book 61 (p. 96). 
Because of this any projection of peralka- 
line liquids, either natural or experi- 
mental, into Petrogeny's Residua system 
for purposes other than simple classifica- 
tion may be misleading. The situation 
becomes much worse when the normative 
calculation forms the basis of such a 
projection — a procedure that has been 
used in several recent discussions and 
whose usefulness has been advocated by 
some authors (Carmichael and Mac- 
Kenzie, 1963, pp. 384-385; Wyllie, 1963, 
p. 65). There are several serious deficien- 
cies in a norm projection for peralkaline 
compositions, which we have had cause 
to note elsewhere in calling attention to 
an unexpected relationship between per- 
alkaline liquids and their feldspar pheno- 
crysts (Bailey and Schairer, 1964a). We 
find that in the available analyses of 
alkali feldspar phenocrysts and parent 
peralkaline rocks the K 2 0/Na 2 ratio is 
higher in the feldspar phenocrysts even 
when the parent material is strongly 
sodic. Separation of such feldspar is a 
fractionation of potash, and we have 
referred to this process as the "orthoclase 
effect" in analogy with the "plagioclase 
effect" (Bowen, 1945), which is a power- 
ful means of fractionating alumina and 
generating peralkaline liquids. The ortho- 
clase effect is masked by the norm 
projection of peralkaline liquids into 
Petrogeny's Residua system; in fact, this 
projection appears to show the converse. 
Because of this Chayes and Metais have 
suggested, on page 193 of this report, an 
alternative to the standard normative 
calculation to cope with the alkali balance 
in peralkaline rocks. Although it will be 

hard to obviate all the difficulties inherent 
in any arbitrary representation of a rock 
composition, one thing is plain: if an 
oxide, e.g., Na 2 0, is partitioned in the 
calculation, and part of it is subsequently 
ignored in a projection, any deductions 
about differentiation or petrogenesis in- 
volving the oxide must be suspect. 

Sodium Silicates 

Determinations of the eutectic temper- 
atures A and B (fig. 10) required X-ray 
detection of the first appearance of sodium 
silicate in the appropriate compositions 
because this was the last phase to crystal- 
lize and was difficult to detect optically. 
It may be noted, in passing, that sodium 
metasilicate is not compatible with 
acmite, albite, or quartz, which is further 
good reason for using the CIPW norm 
only as a classification device, as was its 
original purpose. Sodium disilicate is the 
common sodium silicate in compositions 
of petrological interest, and this apparent- 
ly simple compound raises problems of its 
own. It forms a large and as yet unde- 
termined number of polymorphs, some of 
which were discussed by Kracek, Morey, 
and England {Year Book 52). Mixing of 
polymorphs, strong degrees of preferred 
orientation in some of them, and diffi- 
culty in grinding for X ray complicate the 
issue, but we have encountered two 
predominant forms in our study. Addi- 
tional problems stem from the very rapid 
and frequently metastable crystallization 
of sodium disilicate in some compositions. 
For instance, reexamination of compo- 
sitions near the ternary eutectic acmite + 
albite + sodium disilicate + liquid, and 
on the join albite-sodium disilicate, 
showed the existence of metastable solid 
solutions of albite in sodium disilicate. 

A possibly more disturbing result of 
sodium disilicate's tendency to rapid 
metastable crystallization is that it 
appears to have concealed the existence 
of another sodium silicate between disili- 
cate and quartz. In our study of compo- 
sitions leading to eutectic B, quartz + 
albite + acmite + sodium disilicate + 
liquid, we find that the usual forms of 



sodium disilicate are not present in 
compositions that have been slowly 
crystallized over a long period of time. 
Instead, a phase that has been assumed 
to be another disilicate polymorph ap- 
pears. This phase has not been described 
before; it has a distinctive X-ray pattern 
and was found in the system Na 2 0-MgO- 
Si0 2 (Schairer, Yoder, and Keene, Year 
Book 53), where it was also taken to be a 
form of sodium disilicate (Yoder, personal 
communication). It was labeled W to 
distinguish it from the other new com- 
pounds encountered in Na 2 0-MgO-Si0 2 . 
In our compositions, however, there 
appeared to be discrepancies in the quartz 
content when phase W was present, and 
we are currently finding similar results in 
a reexamination of the join Na 2 0-Si0 2 . 
There is no record of a silicate between 
disilicate and quartz in the system 
Na 2 0-Si0 2 (Morey and Bowen, 1924; 
Kracek, 1930, 1939), but present indica- 
tions are that IF is a new compound of, 
or close to, the composition 3Na 2 • 8Si0 2 . 
The existence of such a compound makes 
little difference to the petrological infer- 
ences to be drawn from Na 2 0-Al 2 3 - 
Fe 2 3 -Si0 2 , for, like sodium disilicate, it 
has yet to be recorded from rocks. The im- 
portance of the sodium silicate-rich eutec- 
tics in the system is that they illustrate 
the crystallization trends of peralkaline 
nepheline syenites and granites. The 
discovery of this new compound will 
necessitate the inclusion of a very small 
liquidus field in each of the ternary 
diagrams Na 2 0-Al 2 3 -Si0 2 and Na 2 0- 
Fe 2 3 -Si0 2 , with an appropriate volume 
in the quaternary system and a quater- 
nary point between A and B (fig. 10), 
with albite -f- acmite -f- sodium disilicate 
+ W + liquid in equilibrium at the new 

Temperature and Vapor Composition 

in Carbonatite and Kimberlite 

D. K. Bailey 

Phlogopite, a characteristic mineral of 
many carbonatites, is abundant in kim- 
berlites, many of which also contain 
dolomitic carbonates. A knowledge of the 

stability of this mineral in a carbonate 
assemblage is therefore necessary to the 
proper understanding of its formation in 
these rocks as well as in the more common 
assemblages of metamorphic limestones. 
Furthermore, some carbonatites and kim- 
berlites contain potash feldspar, in addi- 
tion to phlogopite, so that the relative 
stability of the two potash minerals in a 
carbonatic assemblage should prescribe 
the conditions of formation in these rocks. 
The reaction 

K 2 0-Al 2 3 -6Si0 2 + 


6(Ca0-MgO2C0 2 )+ 2H 2 

Dolomite Water 

i=>K 2 0-6MgO-Al 2 3 -6Si0 2 -2H 2 + 


6(CaO-C0 2 ) + 6C0 2 

Calcite Carbon dioxide 

was suggested by H. S. Yoder, Jr. 
(personal communication) as the simplest 
expression of the possible relationships 
that might exist among these minerals. 
Reactions of this type, decarbonation <=± 
dehydration, though they must be com- 
mon in nature, have not previously been 
studied systematically, but a theoretical 
discussion of various types of reactions 
involving the vapor phase was given by 
Greenwood {Year Book 61, pp. 82-85). 
The equation for the slope of the equi- 
librium boundary curve for this kind of 
reaction, when vapor composition is 
plotted against temperature, at constant 
pressure, was derived by Greenwood as: 

+ oo > 

\dxjp > AS \xj 

where x 2 equals the mole fraction of the 
second component in the binary fluid 
phase. This equation implies the tendency 
to a steep slope eventually approaching 
zero degrees Kelvin for a pure H 2 
atmosphere, and, as Greenwood pointed 
out, such reactions may be particularly 
valuable in defining conditions of forma- 
tion of natural assemblages. 

In the current investigation, pure 
natural orthoclase (Fianarantsoa, Mada- 



gascar) and dolomite (Oberdorf, Austria), 
finely ground and mixed in the propor- 
tions 1:6, are sealed with water in 
platinum capsules, and heated in cold- 
seal pressure vessels. Control of the vapor 
composition has been imposed by weigh- 
ing in known amounts of water to the 
charge. The position of the reaction curve 
is then located by finding the temperature 
at which complete reaction takes place 
for a specific ratio of dry charge to water, 
for if reaction is complete the amount of 
water consumed and the amount of 
carbon dioxide generated can be calcu- 
lated, giving the resultant vapor compo- 
sition. This technique cannot, of course, 
locate the curve as precisely as should be 
possible with an external control of the 
atmosphere, but as the curve is not too 
steep over most of the range the degree of 
inaccuracy will not be large. 

The results to date, plotted as temper- 
ature against vapor composition, for 1 kb 
total pressure, are shown in figure 13, and 






H 2 

30 40 50 60 70 

Mol per cent 

Fig. 13. Equilibrium diagram for the reac- 
tion orthoclase + dolomite + water ?=* phlo- 
gopite + calcite + carbon dioxide at 1 kb total 
pressure. Broken lines intersecting curve depict 
runs in which all four solid phases are present. 
Where water was sufficient for complete reaction 
the maximum content of C0 2 that might have 
been generated is shown by the solid squares. 

the probable limits of error are indicated. 
It will be noted that, in runs in which all 
four solid phases persist, only an upper 
limit for the mole fraction of C0 2 can be 
fixed, although with experience some 
estimate of the extent of reaction can be 
made from the relative proportions of the 
four phases. At the higher temperatures 
reaction is rapid, being complete in a few 
hours; at lower temperatures, although 
breakdown of dolomite is still rapid, the 
formation of phlogopite is sluggish, and 
the curve in figure 13 may not be limited 
so much by the vapor composition as it 
approaches pure H 2 as by the lower 
temperature limit for the formation of 
phlogopite. Variation of the equilibrium 
with change of total pressure is also being 
examined, and the results will be pub- 
lished in a later report. The temperature 
of reaction in a C0 2 -rich atmosphere 
rises with increasing pressure and will 
presumably be limited by eventual inter- 
sections with other breakdown curves for 
dolomite. The marked inflection in the 
boundary curve in figure 13 was unex- 
pected, and it is therefore a pleasure to 
record that an inflection in the curve 
describing this type of reaction has been 
deduced on theoretical grounds by H. J. 
Greenwood (personal communication), 
who expects to publish his general dis- 
cussion in the near future. 

In the dissected carbonatite volcanoes 
of the Rufunsa province in Northern 
Rhodesia there has been extensive potash 
metasomatism with the formation of 
orthoclase rock from a variety of country 
rocks. Some of the orthoclase incorpo- 
rated in the intrusions is rimmed by 
phlogopite, and the carbonatite volcanics 
variously either contain feldspathic frag- 
ments or are rich in phlogopite. It is clear 
from this that initially orthoclase was 
stable, setting an upper temperature 
limit of approximately 600°C to this 
near-surface activity. Subsequently, with 
rising temperature or, more likely, de- 
clining partial pressure of C0 2 , phlogopite 
became the stable potash silicate in 
equilibrium with the carbonates, setting 



a lower limit of temperature around 

Reaction of early-formed orthoclase to 
produce phlogopite is shown schemati- 
cally in figure 14, from which it is readily 
seen that a variety of changes could give 
similar results, within the temperature 
span of the reaction. In the example cited 
a change such as that labeled (3) in figure 
14 would appear the most reasonable 
because the formation of phlogopite rims 
on the orthoclase is plainly a late reaction, 
which would coincide with cooling, and 
with diminishing C0 2 content in the 
vapor, after precipitation of carbonates 
from the late-stage fluids. This tempera- 
ture range, and the inferences about the 
sequence of activity with diminishing C0 2 
pressure, accord with earlier deductions 
based on the field relations and petrog- 
raphy (Bailey, 1960, 19646). 

A survey of the literature reveals that 
potash feldspar rimmed with phlogopite 
has been recorded in similar subvolcanic 
complexes in Nyasaland, Tanganyika, 
and elsewhere, and in other complexes the 
assemblage potash feldspar + dolomite 
frequently occurs in late-stage veins, 













Mol per cent 


Fig. 14. Schematic diagram showing possible 
paths by which orthoclase could react to yield 
phlogopite: (1) with increasing temperature and 
constant vapor composition, (2) with diminishing 
C0 2 content in vapor at constant temperature, 
and (3) with decreasing C0 2 and cooling. 

emphasizing the possible applications of 
this reaction. In older and deeper com- 
plexes such as Nkumbwa Hill, Northern 
Rhodesia, and Phalabora in the Transvaal 
there are zones of phlogopite rock in the 
country rocks adjacent to the carbonatite 
stem, in contrast to the orthoclase rock 
of higher exposure levels such as Rufunsa. 
This is probably a reflection of the ex- 
pected higher temperature with greater 
depth in a complex, whereby a tempera- 
ture would eventually be reached at 
which potash feldspar would be unstable 
in the presence of dolomite, regardless of 
CO 2 pressure (see path 1 in fig. 14). 

Dolomitic carbonates are common in 
kimberlite, and this association together 
with the ubiquitous phlogopite of this 
rock imposes similar limits to those out- 
lined for carbonatites. The initial forma- 
tion of phlogopite would be above 600°C, 
assuming high Pco 2 , but the presence of 
xenoliths of feldspathic granulite, granite, 
or arkose, containing unaltered potash 
feldspar, implies temperatures below the 
range 600°-300°C for the final emplace- 
ment of kimberlite breccia. Such temper- 
atures would be consistent with the 
conspicuous lack of thermal metamor- 
phism of the wall rocks and xenoliths of 
most kimberlites. 

The presence of potash feldspars and 
phlogopite in carbonatitic assemblages 
may therefore be an index of the condi- 
tions of formation; in addition to its 
application in metamorphic petrology, 
this relationship provides useful new 
limits for, and hence better understanding 
of, the conditions of carbonatite and 
kimberlite eruption in the upper levels of 
the crust. 

Breakdown of Monticellite and 

Akermanite at High Pressures 

/. Kushiro and H. S. Yoder, Jr. 

Stability of monticellite (CaMgSi0 4 ) 
and akermanite (Ca 2 MgSi 2 7 ) has been 
examined at high pressures and high 
temperatures. In the present experiments, 
a composition diopside 20 and monti- 
cellite 80 weight per cent was studied over 



a wide pressure and temperature range 
(fig. 15). The solid-media, high-pressure 
apparatus of the piston-cylinder type 
similar to that designed by Boyd and 
England (Year Books 57 and 60) was 
used for most of the runs, and the gas- 
media, high-pressure apparatus designed 
by Yoder (1950) was used for the several 
runs at pressures lower than 10 kb. 

At 1 atm below 1430°C the crystallized 
glass of this composition consists of 
forsterite and monticellite solid solutions 
and pure akermanite. Above 1430°C and 
below about 1450°C the mixture consists 

of forsterite and monticellite solid solu- 
tions and liquid, and above 1450°C and 
up to the liquidus temperature, 1498°C, 
it consists of monticellite solid solution 
and liquid. These experiments were con- 
ducted by Schairer. 

The assemblage monticellite + 
forsterite ss + akermanite is stable up to 
8 kb at 1350° and 1400°C. At 9 kb and 
1400°C monticellite disappears and mer- 
winite (Ca 3 MgSi 2 8 ) appears. The break- 
down of monticellite, therefore, takes 
place between 8 and 9 kb at 1400°C. The 
line A of figure 15 would be close to the 

10 15 

Pressure , kb 

Fig. 15. Pressure-temperature plane for monticellite (CaMgSi0 4 ) 80, diopside (CaMgSi 2 6 ) 20, 
weight per cent composition. Ak, akermanite (Ca 2 MgSi 2 7 ); Di, diopside; En, enstatite (MgSi0 3 ); 
Fo S8 , forsterite solid solution; L, liquid; La, larnite (Ca 2 Si0 4 ); Mer, merwinite (Ca 3 MgSi 2 8 ) ; 
Mo B8 , monticellite solid solution; Ra, rankinite (Ca 3 Si 2 7 ); Wo, wollastonite (CaSi0 3 ). Small squares 
indicate the runs made by gas-media high-pressure apparatus. Lower diagrams show the tie lines 
of coexisting phases between the joins larnite-forsterite and wollastonite-enstatite for different PT 
conditions. Cross indicates the composition studied. 



univariant curve for the breakdown of 
monticellite according to the reaction 3 
monticellite = merwinite + forsterite. 
The volume change in this reaction, from 
left to right, is —7.9 cc. The line A may 
have a zone, since monticellite in this 
system is not pure monticellite but a solid 
solution involving a small amount of 
forsterite. The tie lines of coexisting 
phases between the joins larnite (Ca 2 - 
Si0 4 ) -forsterite (Mg 2 Si0 4 ) and wollaston- 
ite (CaSi0 3 )-enstatite_.(MgSi0 3 ) on the 
lower pressure side of the line A are 
shown in the lower left diagram of figure 
15. Some of them are based on the data 
at 1 atm by Ricker and Osborn (1954). 

The assemblage forsterite ss + mer- 
winite + akermanite is stable up to the 
line B. The forsterite solid solutions show 
larger d values than those of pure 
forsterite, containing most probably a 
small amount of Ca replacing Mg. The 
tie lines of coexisting phases in the 
pressure-temperature conditions between 
the lines A and B are shown in the lower 
center diagram. The stability of rankinite 
and the tie lines including larnite and 
rankinite have not been determined, and 
there may be intermediate stages of the 
tie lines between the lower left and the 
lower center diagrams of figure 15. 

On the higher pressure side of the line 
B, akermanite breaks down into diopside 
and merwinite and the assemblage is 
merwinite + forsterite S9 + diopside. This 
assemblage is stable at least up to 38 kb 
at 1500°C. On the higher pressure side, 
but close to the line B, a small amount of 
akermanite is still left, and the assemblage 
is merwinite + forsterite ss -f- diopside + 
akermanite. However, the presence of 
diopside indicates that akermanite has 
begun to break down into merwinite and 
diopside and that akermanite is not stable 
on the higher pressure side of the line B, 
which is, therefore, a univariant curve for 
the breakdown of akermanite into mer- 
winite and diopside. It was checked by 
using pure akermanite (this report, p. 
85). The tie lines of coexisting phases 
are shown in the lower right diagram of 

figure 15. Experiments on a composition 
on the join Wo-Ak indicate that mer- 
winite coexists with wollastonite at 20 kb. 
Therefore, it is most likely that the 
Mer-Wo tie line is valid on the higher 
pressure side of line B. Between the lower 
center and the lower right diagrams of 
the figure there may be intermediate 
stages of the tie lines, in which La-Ak and 
Wo-Ak tie lines, or Wo-Ak, Ra-Ak, and 
Ra-Mer tie lines, or Ra-Mer and 
Mer-Wo tie lines may exist. 

Monticellite occurs in some meta- 
morphosed dolomitic limestone, igneous 
rocks, and ultramafic rocks as described 
by several authors (e.g., Tilley, 1947; 
Bowen, 1922; and Buie, 1941). The 
results of the present experiments would 
give the upper limits of the formation of 
these rocks. By extrapolating the line A 
to the probable geothermal gradient 
curve within the earth, it is suggested 
that monticellite is not stable below about 
16 km. However, the pressure of the 
stability field of monticellite increases 
with increase of temperature, and, for 
example, at 1250°C, the temperature of 
some basic magmas, monticellite would 
be stable up to about 8 kb, corresponding 
to the pressure at the depth of about 26 
km. It is concluded from the present 
experiments that the monticellite-bearing 
rocks whose formation is related to basic 
magmas would have formed at depths 
less than about 26 km and those related 
to the more salic magmas would have 
formed at depths much shallower than 
this. Stability of akermanite is discussed 
further in another part of this report 
(p. 84). 



I. Kushiro 

The wollastonite (j8CaSi0 3 )-pseudo- 
wollastonite (aCaSi0 3 ) inversion has been 
studied by Osborn and Schairer (1941) at 
1 atm. The inversion temperature is 
1125° db 10°C for pure CaSi0 3 . In the 
present experiments the inversion tem- 
perature and the melting curve of pseu- 



dowollastonite were determined up to 23 
kb. The pressure-temperature plane for 
CaSi0 3 composition is shown in figure 16. 
Point A is the melting point of pseudo- 
wollastonite (1544°C), and B is the 
temperature of the inversion ; both points 
were determined at 1 atm by Osborn and 
Schairer (1941), who indicated that the 
inversion is very sluggish at 1 atm, 4 
weeks being required for complete inver- 
sion. At high pressures, however, the 
inversion takes place rapidly since the 
inversion temperatures are high as com- 
pared with that at 1 atm. For example, 
at 15 kb the inversion of wollastonite to 
pseudowollastonite is completed within 17 
minutes at 1470°C and within 8 minutes 
at 1580°C, and at 20 kb the inversion is 
completed within 6 minutes at 1600°C. 
The reverse inversion, from pseudo- 
wollastonite to wollastonite, also takes 
place rapidly. 

The inversion curve has a slope of 
21°/kb and can be approximately repre- 
sented by T = 1125 + 21P, where T is 




i 1 i 


a a/" 


o m 















/% Wo 


D Melted 
a Wo^Pwo 
B Pwo->Wo 
■ Wo unchanged 




l l i 

10 20 

Pressure t kb 

Fig. 16. Pressure-temperature plane for 
CaSi0 3 composition. L, liquid; Pwo, pseudowol- 
lastonite; Wo, wollastonite. Points A and B 
are from Osborn and Schairer (1941). 

in degrees Centigrade and P in kilobars. 
The inversion curve can be calculated 
from the thermochemical data of wollas- 
tonite and pseudowollastonite. The heat 
of inversion at 25°C is 1.56 kcal/mole, a 
value obtained from the standard heats 
of formation of wollastonite and pseudo- 
wollastonite given by Kracek and co- 
workers (Year Book 52, p. 74). The 
volume difference was found to be 0.15 
cc/mole by using densities of these min- 
erals given by Winchell (1933). The slope 
(dT/dP) calculated from the above data 
is 3.3°C/kb, a much smaller value than is 
obtained from the experiments. The 
difference could be explained by the errors 
of the thermochemical data used in this 
calculation or by the differences between 
thermochemical data at 25°C at 1 atm 
and those at high temperatures and high 
pressures. The experimental data are 
more reliable than the calculation, and 
therefore the thermochemical data for the 
inversion at high pressures and tempera- 
tures can be estimated by the slope 
obtained by the experiments. If the 
volume change is 0.15 cc/mole the heat 
of inversion is approximately 9.9 kcal/ 
mole, and if the heat of inversion is 1.56 
kcal/mole the volume change is about 
0.95 cc/mole. 

The melting curve of pseudowollaston- 
ite has a slope of 3°/kb and intersects 
the inversion curve at about 23 kb and 
about 1610°C. Pseudowollastonite is, 
therefore, not stable higher than 23 kb at 
any temperature. As shown in figure 16 
the stability field of pseudowollastonite is 
limited to the very high temperature 
region as compared with the temperatures 
of formation of most rocks, except for 
some basic igneous rocks. 

Stability Field of Akermanite 
I. Kushiro 

Akermanite (Ca2MgSi 2 7 ) is one of the 
main components of natural melilites, and 
its stability field would give an indication 
of the physical conditions of formation of 
the melilite-bearing rocks. In the present 
experiments the stability field of aker- 



manite has been determined at high 
pressures and high temperatures. As 
shown in figure 17 the univariant curve 
for the breakdown of akermanite into 
merwinite and diopside, the same as the 
line B of figure 15, was determined by 
using synthetic pure akermanite and a 
partially crystallized glass having the 
composition diopside 20 and monticellite 
80 weight per cent. On the higher pressure 
side, but close to the curve, a small 
amount of akermanite is still left. How- 
ever, presence of merwinite and diopside 
indicates that akermanite is not stable on 
the higher pressure side of the curve. The 
duration of the runs was about 1 hour 
near 1300°C and about 30 minutes near 
1400°C ; however, the runs might not have 

been long enough for complete reaction. 
At 900°C the reaction was completed 
within 71 hours at 9.1 kb. 

Akermanite also breaks down into 
monticellite and wollastonite below 700°C 
at pressures lower than about 4 kb 
(Harker and Tuttle, 1956). The slope 
(dT/dP) of the univariant curve is almost 
zero, and therefore this reaction is 
extremely temperature sensitive. The 
univariant curve of this reaction may be 
cut by the univariant curve of the break- 
down of monticellite into merwinite and 
forsterite (line A of fig. 15) at about 5 kb. 
Therefore the assemblage merwinite + 
forsterite 3S + wollastonite may exist be- 
tween the assemblages monticellite + 
wollastonite and merwinite + diopside. 






/ Mer 

Mo+Wo / * 
/ Fo 

? ' Wo 

; i 


/ Mer + Di 

Ak composition 
□ Gl 

[I Mer + GI 

3 Mer + Di + Ak 
M Mer + Di 

Di2o Moeo composition 

CD Ak + Mer+ft>ss 
O Mer + Di + Fo ss + 


10 I! 

Pressure , kb 


Fig. 17. Pressure-temperature plane for akermanite (Ca 2 MgSi 2 07) composition. Ak, akermanite; 
Di, diopside (CaMgSi 2 6 ); Fo sa , forsterite solid solution; L, liquid; Mer, merwinite (Ca 3 MgSi 2 8 ); 
Mo, monticellite (CaMgSi0 4 ); Wo, wollastonite (CaSi0 3 ). Points A and B are from Osborn and 
Schairer (1941) and Harker and Tuttle (1956), respectively. 



The volume relations among the four 
different assemblages of akermanite com- 
position are as follows : 

akermanite = monticellite + 

(A7= -2.2 cc) 

wollastonite = Y» merwinite + 

(A7= -2.6 cc) 

Y forsterite + wollastonite 
= Yi merwinite + H diopside 

(AV = -5.2 cc) 

It is indicated from the volume relations 
that merwinite + diopside is the densest 
assemblage of all, and the volume differ- 
ence between akermanite and merwinite 
+ diopside is 10 cc per akermanite mole. 

The univariant curve of the reaction 
akermanite = Yi merwinite + Yi diop- 
side can be calculated from the volume 
change and the standard heat of reaction 
given by Neuvonen (1952) on the 
assumption that the entropy and volume 
changes of the reaction are almost the 
same over a wide temperature and pres- 
sure range. The standard heat of reaction 
given by Neuvonen is -2300 ± 300 
cal/mole. If the entropy change of the 
reaction is —3.4 cal/deg«mole, that is, 
the extension of the univariant curve 
intersects the temperature axis at 400°C, 
the slope (dT/dP) of the univariant curve 
is 0.071 deg/bar and the equation of the 
curve is T = 400 + 0.07LP, where T is 
in degrees Centigrade and P in bars. This 
curve passes 1110°C at 10 kb, and 1465°C 
at 15 kb, and is close to the curve ob- 
tained by the present experiments, indi- 
cating that the experimental results are 
not inconsistent with the thermochemical 

Akermanite melts incongruently to 
produce merwinite and liquid at high 
pressures ; the liquid thus produced is rich 
in diopside. The pressure at which aker- 
manite begins to melt incongruently 
would be about 5 kb, although it was not 
determined exactly. 

The stability field of akermanite shown 
in figure 17 indicates that akermanite is 
stable at temperatures higher than 700°C 
but lower than 1500°C and at pressures 

lower than 15 kb. The geothermal 
gradient curve is most unlikely to pass 
the stability field of akermanite, and, 
therefore, akermanite would not be stable 
in most conditions within the earth 
except for the unusually high temperature 
conditions such as those in magmas 
brought up to the shallower part of the 
earth's crust. This is consistent with the 
observations that akermanite and meli- 
lites rich in the akermanite component 
occur in igneous rocks and rocks ther- 
mally metamorphosed by the intrusion of 

Soda Melilite 

H. S. Yoder, Jr. 

The alkali basalts are identified in part 
by the presence of nepheline in their 
norm as indicated by the fundamental 
basalt tetrahedron of Yoder and Tilley 
(1962). The basalt tetrahedron has now 
been expanded to include normative 
larnite (see fig. 1) for the purpose of 
illustrating principally the melilites aker- 
manite and soda melilite, because melilite 
solid solutions involving these molecules, 
Ca 2 MgSi 2 07 and CaNaAlSi 2 7 , respec- 
tively, are in intimate association with 
nepheline in some alkali rocks. 

The compositions of melilites occurring 
in igneous rocks are represented for the 
most part in the system akermanite-soda 
melilite-gehlenite (fig. 18). A significant 
number of melilites plot near the middle 
portion of the soda melilite-akermanite 
join, even though emphasis in the past 
has been directed toward the gehlenite- 
akermanite join. Natural melilites close to 
soda melilite are not known, and attempts 
to synthesize soda melilite at 1 atm have 
not been successful (W. R. Foster, 1942; 
Schairer and Yoder, unpublished data, 
1964). 5 The equivalent bulk composition 
is represented by nepheline + wollaston- 

CaNaAlSi 2 7 -> NaAlSi0 4 + CaSi0 3 

6 The alleged synthesis of soda melilite by- 
Nurse and Midgley (1953) was not found accept- 
able (see Yoder and Tilley, 1962, p. 356; Christie, 
1962, p. 10). 




Ca 2 AI 2 SiO- / 

Mol per cent 

Ca 2 MgSi 2 7 

Fig. 18. Plot of natural melilites in terms of the molecules soda melilite (CaNaAlSi 2 7 ), gehlenite 
(Ca2Al 2 Si07), and akermanite (Ca2MgSi 2 07). Analyses reduced after converting to atoms on a 
14-oxygen basis by first assigning sufficient Al + Fe +3 to fill the silicon position and making equiv- 
alent gehlenite; Fe +2 + Mg + Mn = akermanite, and Na + K = soda melilite, which are then 
adjusted to 100 mole per cent. Filled circles, Sahama and Meyer (1958); half solid circles, Neuvonen 
(1955); open circles, Buddington (1922); circle and cross, Tilley and Henry (1953); open square, 
Washington (1927); half solid square, Larsen (1942); square and cross, Ramsay (1921); open tri- 
angle, Tilley (1929). Dashed line is estimate of limit of synthetic melilite solid solutions at the solidus 
at 1 atm pressure based on the work of Goldsmith (1948) and Schairer and Yoder (unpublished 
data, 1964). Molecular weights of end members are soda melilite, 258.16; gehlenite, 274.16; and 
akermanite, 272.60. 

That assemblage is known in the wollas- 
tonite ijolite dikes and metamorphosed 
blocks. Because of the importance of the 
soda melilite molecule in natural melilites 
and its role in the fractionation of alkali 
basalts, efforts were made to outline the 
field of stability of soda melilite. 

The preliminary results shown in figure 

19 outline the field of stability of soda 
melilite with the following reservations. 
First, the pressures marking the above 
reaction may not be the lowest because of 
the sluggish nature of the reaction. The 
curve as drawn represents the lower 
pressure limit for anhydrous runs up to 
2 weeks in duration. Hydrothermal exper- 



1 0,000 1 1 1 r 





In 5000 



i2 4000 





Soda Melilite 

Sm + Wo + LH^/f 

' m 

Sm + Ne ss +Wo + L \ /?'/ 




Ne ss + Wo+L 

Nepheline ss 4- Wollastonite 

Ness + Pwo + L- 

-Pwo + L 










Temperature ,° C 

Fig. 19. Preliminary results outlining the stability field of soda melilite (NaCaAlSi 2 7 ). Cross, 
soda melilite; cross in square, soda melilite (Sm) + liquid (L); circle, all liquid; triangle, nepheline 
solid solution (Ne S8 ) + wollastonite (Wo); diagonally filled square, soda melilite + nepheline solid 
solution -f- wollastonite + liquid; half filled circle, soda melilite + pseudo wollastonite (Pwo) + 
liquid; triangle with cross, nepheline solid solution + pseudo wollastonite + liquid; circle with cross, 
pseudowollastonite + liquid. Amount and kind of solid solution in soda melilite have not been 
determined. Negative slope of solid state reaction probably due to nonequilibrium. 



iments are now under way in an effort to 
increase the reaction rate. The negative 
slope of the solid state reaction is most 
likely the result of decreasing reaction 
rate rather than volume or entropy 
changes. Second, the melting of soda 
melilite takes place over a range of 
approximately 50°C, from which it may 
be concluded that the melilite crystal- 
lizing from liquid is not CaNaAlSi 2 7 
exactly but a complex solid solution. The 
temperature of complete melting of a 
soda-rich melilite is given by the highest 
temperature curve. 

The results shown at 1 atm are from 
Schairer and Yoder (1964, unpublished 
data; cf. fig. 3). They are not in agreement 
with the sequence of events that can be 
deduced from the relations between 
nepheline and wollastonite as determined 
by Foster (1942, p. 160). He shows a 
eutectic relationship between nepheline 
and pseudo wollastonite, where Schairer 
and Yoder found nepheline and wollaston- 
ite or pseudowollastonite in equilibrium 
with liquid over a range of about 40°. 
It appears that nepheline-wollastonite is 
pseudobinary ; however, the nature of the 
complex solid solutions involved is not 
known. Some of the complexity may be 
due to the extensive metastability occur- 
ring in nepheline solid solutions. 6 

The slope of the Wo ss -Pwo ss inversion 
is about 13°/kb, which is much less than 
the rate given by Kushiro (this report) 
for pure CaSi0 3 , 21°/kb. Metastable 
formation of pseudowollastonite from 
liquid would account for the discrepancy, 
although the two runs fixing the maxi- 
mum slope are based on the conversion of 
wollastonite solid solution. 

It appears that soda melilite becomes 
stable at relatively low pressures and is, 
therefore, to be expected as a common 
molecule in melilite solid solutions formed 

6 Schairer and Yoder (1962, unpublished data) 
have found nepheline solid solutions close to 
jadeite in composition which with very long runs 
slowly exsolve to nepheline solid solution and 

in the crust. Associations of nepheline and 
wollastonite should be restricted to rela- 
tively low pressure conditions, e.g., 
wollastonite nephelinite lavas, wollaston- 
ite ijolitic dikes, and some contact 
metamorphic aureoles. 

The Join Akermanite 
(Ca 2 MgSi 2 7 )-SoDA Melilite 

(NaCaAlSi 2 7 ) 

J. F. Schairer and H. S. Yoder, Jr. 

Many analyses of melilites separated 
from igneous rocks indicate that up to 
about 44 weight per cent of the soda 
melilite molecule (NaCaAlSi 2 C>7) is pres- 
ent in some melilites. Compositions 
between akermanite and soda melilite lie 
in the join Ne-Ak-CaSi0 3 (fig. 3). Four- 
teen compositions between akermanite 
and soda melilite were prepared and 
studied. The join is not even approxi- 
mately binary at any temperature at 1 
atm pressure. The exact composition of 
none of the phases except the pure 
akermanite composition itself is known. 
The preliminary data obtained are given 
here in figure 20. Not all the data are 
perfectly consistent, and near the soda 
melilite side X-ray and microscopic data 
were inadequate to identify the nature 
and number of the phases present at some 
temperatures. The preliminary diagram 
of this join is given here because it is so 
widely at variance with the data previ- 
ously given by Nurse and Midgley (1953). 

Attention is called to the differences in 
the temperature of appearance of either 
wollastonite solid solution or pseudo- 
wollastonite as a second solid phase after 
melilite has begun to crystallize. The 
small temperature interval where both 
wollastonite and pseudowollastonite ap- 
pear together is not shown in figure 20. 
If the melilite compositions were in the 
join Ca 2 MgSi 2 7 -NaCaAlSi 2 7 this sec- 
ond solid phase should appear at the same 
temperature for all these mixes. The 
problems of synthesis (possible soda 
volatilization) and phase identification 
are now being studied in detail. 







1 1 1 1 1 1 1 1 1 

\ ^~ S1 ^~^~~-~->t-^ LIQUID 

\ ^ _^ 

\ ^ ^^^ 

_ V Mel+Liq 2-^^ 

X \ V \ ~~~~"~^~~^ — .-. ^^^^^ Pwo + Liq^ 

\ ^-^^_ Mel + Pwo+Liq — — ^^^-e J 


^^•^. Mel + Wo+Liq ^-Tr~~~~^~~ = ^~* 

--^ „-»"" Mel + Wo+ Ness+ Liq ^-"*x 

^---^-'--^ — — *" 



Mel + Wo Mel + Wo + Ness 

1 1 1 1 1 ! 1 1 1 

- 1500 


Akermanite 10 
Co 2 Mg Si 2 7 

40 50 60 

Weight per cent 

•90 Soda -Melilite 
Na Co Al SLO, 

Fig. 20. Preliminary phase-equilibria data on the join akermanite-soda melilite. Abbreviations 
as in figures 2 and 3, as well as Pwo, pseudowollastonite. 

The Join Akermanite-Soda Melilite 

I. Kushiro 

Most natural melilites contain soda 
melilite (CaNaAlSi 2 7 ) in solid solution. 
For example, melilites from the melilite- 
bearing rocks of Vesuvius (Buddington, 
1922), Iron Hill (Larsen, 1942; J. V. 
Smith, 1953), Scawt Hill (Tilley, 1929), 
and Hawaii (Neuvonen, 1952) contain at 
least 30 molecular per cent soda melilite, 
and melilite from the nepheline-melilite 
rock of Villa Senni (Washington, 1927) 
contains 43.9 per cent soda melilite. Soda 
melilite is isochemical with nepheline 
(NaAlSi0 4 ) + wollastonite (CaSi0 3 ), 
and, therefore, soda melilite should have 
a different stability field from that of 
nepheline + wollastonite. Yoder (this 

report, pp. 86-89) and Schairer and Yoder 
(pp. 89-90) studied the stability of 
soda melilite, showing that soda melilite 
is unstable at 1 atm and becomes stable 
at pressures higher than about 4 kb. On 
the other hand, akermanite is stable at 
low pressures and breaks down into 
merwinite and diopside at high pressures 
(pp. 84-86). Consequently complete 
solid solution between akermanite and 
soda melilite exists only in the inter- 
mediate pressures. The phase equilibria 
on the join akermanite-soda melilite at 
different pressure is, therefore, of special 

In the present experiments, the phase- 
equilibrium relations on the join aker- 
manite-soda melilite have been deter- 
mined at 20 kb (fig. 21). At this pressure, 
soda melilite is stable, whereas akerman- 






! -■ J / Mel ss -fMerw + Di 

CaNaAiSi 2 7 

20 30 40 50 60 70 80 

Weight per cent Ca 2 MgSi 2 7 


Fig. 21. Equilibrium diagram of the system akermanite (Ca 2 MgSi 2 7 )-soda melilite (CaNaAl- 
Si 2 7 ) at 23 kb pressure. Di, diopside (CaMgSi 2 6 ); L, liquid; Mel ss , melilite solid solution; Merw, 
merwinite (Ca 3 MgSi 2 8 ). 

ite is not stable, and the solid solution 
between them extends from soda melilite 
to about soda melilite 30, akermanite 70, 
weight per cent at temperatures near the 
beginning of melting. In the akermanite- 
rich part of this join, there is a three- 
phase region melilite solid solution + 
diopside + merwinite, since akermanite 
breaks down into diopside and merwinite, 
and the liquidus of merwinite covers 
about 50 per cent of this join. This join 
is, therefore, ternary at 20 kb except the 
join between B and soda melilite, which 
is binary. 

Soda melilite melts congruently at 
1350° ± 15°C at 20 kb, and the tempera- 
tures of both the liquidus and the solidus 
of melilite solid solution increase as the 
akermanite content increases. The tem- 
perature difference between liquidus and 
solidus of soda melilite-akermanite solid 
solution is less than 30°. Since no quench 

crystals were observed in most runs near 
or above the liquidus temperatures, the 
liquidus and the solidus could be deter- 
mined within the error of ±15°. Point A 
is most probably the projection of a 
ternary reaction point, where diopside 
reacts with liquid to produce merwinite 
and melilite solid solution. 

The limit of the solid solution between 
akermanite and soda melilite has not been 
determined exactly, but a charge of 
composition akermanite 70, soda melilite 
30, per cent contains rare merwinite and 
diopside crystals at 1300°C, indicating 
that the limit of the solid solution is close 
to this composition at 1300°C at 20 kb. 

The range of the melilite solid solution 
between akermanite and soda melilite 
would change with pressure. Figure 22 
shows the probable field of the melilite 
solid solution at pressures between 1 atm 
and 20 kb at 1000°C. Akermanite is stable 



1 1 1 [ 

I Melgg 

1 + 

, Merw 

\ + 


\ Diop _ 






1000 °c 


+ \ 

Ne + Wo \ 

i i p \i 



1 1 1 

CaNaAISi 2 7 50 Ca 2 MgSi 2 7 

Molecular per cent 

Fig. 22. Probable field of melilite solid solu- 
tion in the system akermanite (Ca 2 MgSi 2 7 )- 
soda melilite (CaNaAlSi 2 O v ) at 1000°C based on 
data (this report) by Schairer and Yoder; 
Yoder; and Kushiro. Diop, diopside (CaMg- 
Si 2 6 ); Mel ss , melilite solid solution; Merw, 
merwinite (Ca3MgSi 2 8 ); Ne, nepheline (NaAl- 
Si0 4 ); Wo, wollastonite (CaSi0 3 ). 

up to 10.5 kb at 1000°C, which is obtained 
by extrapolating the univariant curve for 
the breakdown of akermanite shown in 
figure 17. On the other hand, soda melilite 
is stable at pressures higher than about 
4 kb at 1000°C (Yoder, this report, p. 
88). At 1 atm, the solid solution extends 
from akermanite to about akermanite 43, 
soda melilite 57, weight per cent, although 
the compositions of the melilite solid 
solution, most probably, do not lie on 
this join (Schairer and Yoder, pp. 
89-90). As shown in the figure, the com- 
plete solid solution between akermanite 
and soda melilite is expected at pressures 
between 4 and 10.5 kb at 1000°C. If the 
temperature is higher than 1000°C, the 
pressure range of the complete solid solu- 
tion would be higher than that at 1000°C. 
It is noticed that the melilite solid 
solutions with compositions between aker- 
manite 40 and 70 molecular per cent are 
stable over a wide pressure range, namely 
from 1 atm to 20 kb. It is suggested from 

these results that melilites rich in aker- 
manite would have formed at relatively 
low pressures, whereas the melilites rich 
in soda melilite could have formed at high 
pressures, even higher than 20 kb. Most 
melilites in lavas contain more than 30 
molecular per cent soda melilite, and 
those with less than 30 per cent of this 
component are rare. This evidence could 
be explained by the results of the present 
experiments. If most melilite-bearing 
lavas originate in the depths near or 
deeper than about 60 km (>20 kb), and 
the melilites crystallize near 60 km, the 
melilites contain more than 30 molecular 
per cent soda melilite as shown in figure 
22. So far as they do not change their 
compositions significantly up to the 
surface, their compositions are limited on 
the more soda melilite rich side of the 
composition soda melilite 30, akermanite 
70, per cent. 

Preliminary experiments on the sta- 
bility of gehlenite indicate that gehlenite 
is stable at least up to 30 kb, suggesting 
that the stability field of melilites would 
be expanded with pressure by the pres- 
ence of gehlenite in solid solution. 

New Relations on Melting 

of Basalts 

C. E. Tilley, H. S. Yoder, Jr., and J. F. Schairer 

Experimental studies on the melting 
relations of a group of basalts under 
anhydrous conditions at 1 atm reported 
in Year Book 62, pages 77-84, have been 
continued to cover the behavior of a 
further group of basalts of contrasted 
composition. Analytical data on these 
assemblages, both of tholeiitic and alkali 
type, are set out in table 1, and the 
results of thermal treatment of these 
assemblages are briefly reported in table 
2. The new determinations provide liqui- 
dus data on rocks ranging in iron enrich- 
ment from 0.366 for an accumulative 
picrite basalt to 0.99 3 for a ferrogabbro 
believed to represent a residual liquid of 
the Skaergaard intrusion layered series — • 
extremes covering a liquidus temperature 
range of 400°C (1420° to 1020°C). 


TABLE 1. Chemical Analyses and Norms of Investigated Rocks 


100.13 100.35 












Sk 3 




Si0 2 









A1 2 3 









Fe 2 3 













































Na 2 









K 2 









H 2 + 








H 2 0" 








Ti0 2 









P 2 5 









Cr 2 3 






























































































MK. Picrite basalt, Kaula Gulch, Mauna Kea, Hawaii (Muir and Tilley, 1963, table 10, analysis 

HK. Ankaramitic picrite basalt, north of Pakaoa Hill, Haleakala, Maui, Hawaiian Islands 
(cf. Macdonald and Powers, 1946, pp. 115-124). Analyst, J. H. Scoon. 

ML 1935. Olivine basalt, 1935 pahoehoe flow of Mauna Loa, Hawaii (Muir and Tilley, 1963, 
table 6). 

El. Olivine basalt, Eldgja-Katla lava, Iceland (Robson and Spector, 1962, p. 1278). Analyst, 
G. R. Robson. 

E2. Olivine basalt, Eldgja-Katla lava, Iceland (Robson and Spector, 1962, p. 1278). Analyst, 
G. R. Robson. 

Sk 3 . Fayalite-ferrogabbro, 2540 meters in layered series, west face of Base Peak, Skaergaard 
intrusion, East Greenland (Wager, 1960, p. 370, table 1, analysis 19). 

LB. "Tholeiite," Cape Vogel area, Papua. Analyst, A. McClure (cf. Joplin, 1963, p. 199, 
analysis 13). 

LBa. Analysis of LB recalculated anhydrous. 

* Loss on ignition. 

f Includes chromite. 



TABLE 2. Results of Thermal Treatment of the Basaltic Rocks of Table 1 
and Others Referred to in Text 

n of 


Picrite basalt, Mauna Kea (MK) 
Picrite basalt, Haleakala (HK) 
Eldgja-Katla olivine basalt (El) 
Eldgja-Katla olivine basalt (E2) 
Ferrogabbro, Skaergaard (Sk 3 ) 
Olivine basalt, Mauna Loa (ML 1935) 
Basalt, Picture Gorge (PG) 
Beaver Bay diabase (Bb) 
Glass ("olivine tholeiite") (G) 

F/(F + M)0.44 2 
"Tholeiite," Cape Vogel area, Papua (LB) 

01 (1420°), Cpx (1210°), PI (1165°) 

01 (1320°), Cpx (1190°), PI (1165°) 

PI, Cpx (1140°); 01 (1120°) 

PI (1120°); Cpx, 01 (1110°) 

PI (1020°); Cpx, FeWo, Fa (995°) 

01, Cpx (1180°); PI (1160°) 

PI, Cpx (1155°) 

PI, Cpx (1085°); 01 (1060°) 

01 (1330°) 
Pr (1385°) 



Cpx, clinopyroxene; Fa, fayalite; FeWo, iron wollastonite; 01, olivine; PI, plagioclase; Pr, pro- 
toenstatite; F/(F + M), iron enrichment for sample G referred to in the text. 

The temperatures are those at which a particular phase disappears on heating long enough to 
reach equilibrium at approximately 1 atm. 

The writers are greatly indebted to the donors of the rocks under study in this account, particu- 
larly to Dr. G. R. Robson, who also supplied the analytical data on the Icelandic lavas, Dr. H. A. 
Powers, Professor L. R, Wager, Dr. G. M. Brown, Professor E. F. Osborn, and Messrs. W. B. Dall- 
witz and J. E. Thompson. 

The observed liquidus of the tholeiitic 
pahoehoe basalt of the 1935 Mauna Loa 
flow, and even of the two transitional 
olivine basalts from Iceland (El, E2), 
falls on or near the linear tholeiitic course 
(Kilauea 1921 lava, iron-enriched schlier, 
K) of figure 23 (fig. 14 of Year Book 62, 
p. 82). The two Icelandic basalts kindly 
provided by Dr. G. R. Robson are 
regarded by Robson and Spector (1962) 
as the extreme members of a fractionation 
series El = (201) - E2 = (21). 

The temperature difference of the two 
liquidi is, however, only 20°C, and that 
of the appearance of the three phases 
(PI, Cpx, 01), 10°C. The meaning of this 
small temperature difference must await 
information on the composition of the 
mineral phases of this suite of rocks. 

Of the Skaergaard assemblages, Sk 2 , 
the liquidus of which was previously 
reported at a temperature of 1035°C, is 
regarded as an accumulate of the upper 
layered series. The specimen Sk 3 (EG 
4330) kindly supplied by Professor L. R. 
Wager and Dr. G. M. Brown is considered 
to represent closely in composition the 

final residual liquid of the Skaergaard 
layered series of which Sk 2 is a cumulus 
(Wager, 1960; Brown and Vincent, 1963). 
It is in conformity with this interpreta- 
tion that the liquidus of Sk 3 (1020°C) is 
lower than that of its cumulus Sk 2 ; the 
temperature difference, however, is only 
15°C (table 2). Held at 975° and 980°C 
for 2 weeks, the coexisting phases are 
plagioclase, clinopyroxene, fayalite, and 
liquid (n of glass 1.530); but at 990°C 
this assemblage is accompanied by iron 
wollastonite (FeWo). Inversion in the 
rock environment is taken at 985°C. 

Turning to the region of lower iron 
enrichment, new data are supplied on a 
glass adjusted to a composition approxi- 
mating an olivine tholeiite which was 
kindly supplied by Dr. D. H. Green. The 
plotted position of the composition in 
figure 23 falls on a linear course of the 
Kilauea 1840 picrite basalt (1840p b ) and 
the 1921 olivine tholeiite (1921). This 
upper linear tholeiitic course of figure 23 
(1840p b -1921) corresponds to a change of 
iron enrichment accompanying the pre- 
cipitation of the dominant phase, olivine, 







Fe0 + Fe 2 3 /Mg0 + Fe0 + Fe 2 3 

Fig. 23. Plot correlating liquidus temperatures with iron enrichment. Numbered and lettered 
points (MK, LB, HK, G, ML 1935, El, E2, PG, Bb, and Sk 3 ) refer to the rocks of tables 1 and 2. 
The remaining points are reproduced from Year Book 62 (p. 82, fig. 14). Temperatures of attainment 
of the four-phase boundary (01, PI, Cpx, liquid) for the respective Hawaiian basalts are marked by 
crosses. The arrowed lines in the diagram plot the iron enrichment at the level of the lower linear 
course of an olivine-normative basalt glass of the 1959 Kilauea Iki eruption (Murata and Richter, 
1964) and of a quartz-normative Pele's hair of the 1921 Kilauea eruption (Tilley, 1960). 

the composition of which becomes less 
forsteritic as crystallization proceeds, as 
is borne out by optical data on the 
olivines of the two assemblages (1840pb, 
1921) which limit the range of the upper 
curve, those of the 1921 lava being the 
less forsteritic (Muir and Tilley, 1963, p. 
116, fig. 3). 

The ankaramitic picrite basalt of 
Haleakala (HK) (Macdonald and Powers, 
1946, p. 115) is an exception, for it 
falls significantly off the upper linear 
tholeiitic course. This rock, however, is a 
picritic rock of alkali type containing 
prominent phenocrysts of titaniferous 
augite (Washington and Merwin, 1922, 
p. 119) as well as olivine. Moreover, 

nepheline as well as minor anorthoclase 
and analcime are constituents of the 
groundmass. The distinctive composition 
of this alkali rock is doubtless the factor 
responsible for its plotted position off the 
main tholeiitic trend in figure 23. The 
Mauna Kea picrite basalt (MK) of less 
markedly alkali type falls near the posi- 
tion of the tholeiitic picrite basalt 
(1840 Pb ). 

The coprecipitation temperature in- 
volving three solid phases, corresponding 
to attainment of the four-phase curve 
(01, PI, Cpx, liquid) in the simple criti- 
cally undersaturated, iron-free basalt 
system (diopside -f orsterite - albite - anor - 
thite, Yoder and Tilley, 1962, p. 395, fig. 



10) has, for a wide variety of Hawaiian 
basalt compositions now investigated, a 
narrow temperature range. Taking the 
variation from 1840 Pb to the 1955 basalt 
(1955-77) where the three phases (01, PI, 
Cpx) coexist at the liquidus (1155°C), the 
range of temperature is only 30°C 
(1140°-1170°), whereas the liquidus tem- 
perature range for the same group of 
basalts is 280°C. These basalts thus ap- 
pear to reach a four-phase boundary 
curve in the more complex natural system 
at about the same temperature and are 
related to one another by the addition or 
subtraction of a phase or assemblage of 
phases, clearly of changing composition, 
as successive solid solutions. 

The least iron-enriched (most basic) 
liquid recognized by chemical analysis on 
the Kilauea suite of lavas is represented 
by a basalt glass of the 1959 Kilauea Iki 
eruption. This activity has been described 
by Richter and Eaton (1960), and a study 
of the chemistry of the lavas has been 
made by Murata and Richter (1964). 

We are grateful to the U. S. Geological 
Survey for permission to refer to the 
basalt glass analysis and that of the 
picrite basalt from which the glass was 
separated. The glass has an iron enrich- 
ment of 0.53 3 , is olivine normative (6.7 
per cent), and its position is indicated in 
figure 23 by an arrowed line across the 
lower linear course at 0.53 3 . The picrite 
basalt (iron enrichment 0.37 8 ) is chem- 
ically almost identical with the 1840 
picrite basalt (0.37 4 ) of our experimental 
runs. It is, as calculation shows, related 
strictly to the glass as an accumulate, 
with olivine as the sole accumulating 
silicate phase. In figure 23 its position, 
plotted on the lower linear course, falls 
close to its intersection with the upper 
course, along which crystallization of 
olivine is dominant. On the present evi- 
dence basalts along the upper course are 
interpreted as strictly accumulative types 
in the petrogenetic sense. 

The temperatures of attainment of the 
four-phase boundary for the series of 
Hawaiian basalts (1840 P b-l 955-77) are 

marked by crosses in figure 23. The two 
high-alumina assemblages, the Skaer- 
gaard and Stillwater chilled margins, 
crystallizing plagioclase on the liquidus 
have a higher temperature for coprecipi- 
tation of three phases, 1190° and 1210°C, 
respectively, the latter rock involving the 
crystallization of an orthopyroxene phase 
(PI, 01, Cpx). For the most iron-enriched 
assemblage studied, the Skaergaard ferro- 
gabbro (Sk 3 ), the corresponding copre- 
cipitation temperature has fallen to 995°C 
with an assemblage that includes fayalite, 
clinopyroxene, and iron wollastonite. 

From a quartz-normative Columbia 
River basalt (PG) from the Picture Gorge 
section, Grant County, Oregon (analysis 
in Hamilton, Burnham, and Osborn, 
1964, p. 23), only two silicate phases (PI, 
Cpx) crystallize. Its liquidus, obtained 
under conditions in which the rock 
powder was sealed in a platinum tube and 
surrounded with iron filings in an evacu- 
ated silica tube, lies well above the linear 
course of figure 23. The liquidus temper- 
ature, 1155°C (PI, Cpx), may be com- 
pared with that recorded by Fudali, 
Muan, and Osborn, 1172°C (plagioclase), 
in the paper just cited (p. 37). Their 
result was obtained at atmospheric pres- 
sure under an /o a slightly higher than that 
of a magnetite-fayalite-quartz buffer, 
stated to be in equilibrium with the 
FeO/Fe 2 3 ratio of the original rock 
(Fe 2 3 4.89, FeO 9.07); however, the 
details of the experiment are not as yet 
available. Note is made of the high ferric 
oxide content of the rock and the presence 
of large amounts of modal magnetite. No 
magnetite was observed in the new runs 
buffered with iron filings. The liquidus of 
the iron-enriched Beaver Bay diabase 
(Bb) of Minnesota (Muir, 1954, p. 377, 
table 1) was obtained under experimental 
conditions similar to those for the Picture 
Gorge basalt. The iron enrichment of this 
rock is identical with that of the iron- 
enriched schlier of Kilauea Caldera (K). 
Plagioclase and clinopyroxene appear 
together at 1085°C, 5° above the liquidus 
of the iron-enriched schlier, and are 



followed by iron-rich olivine at 1060°C. 
Olivine, however, is not present in the 
mode of the schlier, or crystallized in the 
experimental runs. Attention is called to 
the early crystallization of olivine in the 
diabase itself, which is quartz normative. 
If the ferric oxide (4.11 per cent) in the 
rock is reduced to the ferrous state, as 
holds in the experimental run, the compo- 
sition becomes olivine normative (6.3 per 
cent) and the total normative silicate 
content is raised by the additional ferrous 
silicate. It may be expected that the 
equilibrium content of fayalite in the 
olivine eventually precipitated in the run 
is enhanced. The changed ferrous charac- 
ter of the treated rock may provide the 
reason for the reversal of the olivine- 
pyroxene sequence observed in the exper- 
imental run. 

The plot of the "tholeiite" (LB) from 
the Cape Vogel area, Papua, is close to 
that of the Kauai hypersthenic picrite 
basalt (Ka) though its composition is 
strikingly different (table 1, analyses 7 
and 7a). Mr. W. B. Dallwitz and Mr. 
J. E. Thompson of the Bureau of Mineral 
Resources, Canberra, kindly supplied this 
unique material for experimental study, 
and a detailed account of the occurrence 
and petrography of the rock will shortly 
be published by them. As seen from the 
analysis, the rock contains 16 per cent 
MgO and 54 per cent Si0 2 and has 
clinoenstatite and enstatite as prominent 
phenocrysts in a glassy groundmass. The 
clinoenstatite shows evidence that it has 
inverted from a protoenstatite. Thermal 
treatment indicates that protoenstatite is 
the liquidus phase at a temperature of 
1385°C (anhydrous rock, MgO 17.10 per 
cent, Si0 2 57.58 per cent). During the 
course of cooling, protoenstatite contin- 
ues to precipitate, for, when the powdered 
rock is held at 1250°C for 14 days, 
protoenstatite remains the only phase 
crystallizing. When the rock is held at 
1250°C for 6 hours only, enstatite appears 
in abundance as a metastable phase. A 
rock chip heated for 24 hours at 1250° ± 
20°C showed its phenocrysts in large part 

made over to enstatite, but some clino- 
enstatite remained. Long runs are clearly 
required for conversion of clinoenstatite 
through metastable enstatite to proto- 
enstatite. When held at 1150° for 7 days, 
a powdered rock charge showed abundant 
euhedral enstatite accompanied by minor 
clinopyroxene (calcium rich) set in glass. 
The reluctance of clinoenstatite to invert 
to protoenstatite in the absence of a 
liquid phase is shown by the behavior of 
phenocrysts, isolated from the rock, 
subjected to heat treatment. Heated for 
1 hour at 1400°C the material emerges 
unchanged as a white powder, and at 
1200° and 1150°C for 14 days it was still 
almost wholly clinoenstatite with incipi- 
ent change in some grains to protoensta- 
tite. This behavior may be contrasted 
with that of the mineral when heated in 
its liquid environment. 

The protoenstatite developed in these 
charges frequently shows signs of incipient 
inversion to clinoenstatite by the develop- 
ment of the characteristic "crack struc- 
ture" and the resultant onset of twinning. 
Over at least a temperature range of 
135°C protoenstatite thus continues to 
crystallize, whereas with the Hawaiian 
picritic rocks, over an even greater range 
of temperature, olivine is the dominant 
phase precipitated. It is clear that the 
crystallization of a metasilicate in the 
Papuan rock cannot be regarded as the 
metastable equivalent of olivine, as it is 
known to be in experimental work on 
some other compositions (e.g., albite- 
forsterite, this report, p. 98), for the 
bulk composition of the rock has a Si0 2 
percentage of 57.6, with the equivalent of 
more than 12 per cent normative quartz. 
Further experimental studies on this 
pyroxene-rich rock are in progress to 
throw further light on inversion phenom- 
ena in its pyroxene phases. 

Genesis of Principal Basalt Magmas 

H. S. Yoder, Jr. 

In the search for the parental magma 

of basalts it was demonstrated that one 

of the key joins in the synthetic basalt 



system was forsterite-albite (Yoder and 
Tilley, 1962). At 1 atm the join forsterite- 
albite is an equilibrium thermal barrier 
(Schairer and Yoder, Year Book 60, p. 
143), whereas at high pressures (33 kb) 
the barrier is broken by a series of 
pyroxene solid solutions presumably ex- 
tending from enstatite to jadeite. This 
observation along with others led to the 
concept of the shift of thermal barriers 
with pressure and to the possibility that 
a single magma could yield the principal 
basalt (tholeiite and alkali basalt) magma 
types depending on the pressure. 

The extraction in excess of an enstatite- 
rich pyroxene has long been advocated in 
theory as a possible method of deriving 
alkali basalt magma from an olivine 
tholeiite magma crystallizing in the crust 
(Powers, 1935; Macdonald, 1949; Tilley, 
1950) ; however, a detailed analysis of the 
process has not been presented. In terms 
of the forsterite-albite system, it may be 
implied from this hypothesis that a field 
of enstatite appears on the liquidus of the 
join forsterite-albite at pressures less than 
the pressure required to bring about the 
onset of the new thermal barrier enstatite- 
jadeite. (The onset of the enstatite-jadeite 
join obviates the necessity of the enstatite 
extraction method.) The nature of the 
breaching of the thermal barrier by 
pyroxene solid solutions had not been 
investigated in detail by Yoder and 
Tilley (1962), and efforts are now being 
made to ascertain how the encroachment 
of the pyroxene solid solution onto the 
forsterite-albite join takes place at inter- 
mediate pressures. 

The 1 atm diagram for the system 
forsterite-albite is given in figure 24 as 
determined by Schairer and Yoder (Year 
Book 60, p. 143) with a correction of the 
melting point of olivine (1890° ± 20°C). 
Five glasses were prepared in that study 
and crystallized completely at 1050°C. 
One of the crystallized glasses (Fo 3 5Ab65 
weight per cent) was observed to contain 
enstatite as well as forsterite and albite; 
all others contained only forsterite + 
albite. With long runs at temperatures 

immediately below the solidus, the ensta- 
tite in the crystallized starting material 
was observed to disappear. It was con- 
cluded that the enstatite in the starting 
material was metastable even though it 
persisted after 58 days at 1050°C. 

Another example of the metastable 
formation of enstatite was recorded by 
Yoder (1952, p. 588) in a partially 
crystallized glass of the pyrope compo- 
sition. The glass had been held for 3 days 
at 1080°C and consisted of cordierite, 
enstatite, and glass. The stable assem- 
blage was shown to be cordierite + 
forsterite + spinel from the pyrope 
composition and other compositions under 
hydrothermal conditions. Chinner and 
Schairer (1962, p. 616) found that the 
enstatite formed in pyrope glass was 
consumed in long runs under anhydrous 
conditions above 1140°C and the stable 
assemblage was obtained. Below 1140°C 
enstatite persisted in runs of 4 months' 
duration. Hydrothermally the enstatite 
disappeared in 6 to 10 days at 850°C and 
2000 bars, and it was concluded by them 
also that the enstatite was metastable. 

The locations of runs made at 9 kb are 
also shown in figure 24, and a schematic 
interpretation of the results is illustrated. 
The melting point of albite is from the 
work of Boyd and England (1963, p. 318) 
and that of forsterite from Davis and 
England (1964, p. 1115). All the runs in 
which glass was the starting material 
contained large amounts of enstatite 
except those above the inferred liquidus 
and those containing 2 and 3 weight per 
cent forsterite in the initial composition. 
The enstatite appeared as well formed 
rods and cannot be dismissed as a 
quenching product. Below the inferred 
solidus, runs that were initially all 
forsterite + albite recrystallized as such; 
the one composition in which enstatite 
appeared in the initial crystallization 
products remained unchanged in runs up 
to 3 days. Even though the initial starting 
material consisted of forsterite and albite, 
enstatite grew in the liquid produced on 
melting in what is illustrated to be the 




i r 




B6 B20 



20 30 40 50 60 

Weight per cent 





Fig. 24. Forsterite (Fo)-albite (Ab) system at 1 atm of Schairer and Yoder (Year Book 60) with 
correction for melting point of forsterite, and interpretation of preliminary results for the same 
system at 9 kb assuming that enstatite formation is metastable. Compositions studied (B-l, B-6, 
B-20, B-25, B-26) are those of Schairer and Yoder (Year Book 60). 

forsterite -f liquid field. The enstatite is 
believed to be metastable because of the 
inordinate amounts produced at tempera- 
tures immediately below the established 
liquidus and because of its demonstrated 
metastable persistence in much longer 
runs both under hydrous and under 
anhydrous conditions in other compo- 
sitions. The failure of enstatite to grow 
from crystalline forsterite + albite below 
the inferred solidus is not considered 

grounds for concluding that enstatite is 
not stable above the solidus. It could be 
argued that enstatite reacts with liquid 
at an invariant point to produce forsterite 
+ albite as illustrated in figure 25. (The 
known alumina content of enstatite at 
high pressures is neglected for the moment 
to simplify presentation.) This diagram 
has all the requisites in support of 
Powers' (1935) original idea involving the 
extraction of enstatite, but it cannot be 



Weight per cent 

Fig. 25. The nepheline (Ne)-forsterite (Fo)-Si0 2 system at 9 kb predicted from the 1 atm data 
of Schairer and Yoder (Year Book 60) and assuming that the enstatite formed in runs in the Fo-Ab 
system is stable. It is in accord with the requirements of Powers' hypersthene extraction theory for 
the production of alkali basalt magma from an olivine tholeiite magma. Ab, albite; En, enstatite; 
Jd, jadeite; Qz, quartz; Sp, spinel. 

accepted on the basis of the present data 
at 9 kb. This, of course, does not preclude 
the possibility of such thermal relations 
setting in at higher pressures yet below 
the pressure at which forsterite and albite 
become incompatible. The idea that 
settling of excessive metastable hyper- 
sthene from an olivine tholeiitic magma 
will produce the desired nepheline-norma- 
tive liquid is considered improbable. 

Both from laboratory evidence (Boyd 
and England, Year Book 59, p. 49) and 
from the analysis of pyroxenes from 

nodules in alkali basalts, the enstatite 
forming at high pressures is known to 
contain alumina. The alumina is usually 
expressed as the Mg-Tschermak's mole- 
cule. If an enstatite containing a reason- 
able amount of the Mg-Tschermak's 
molecule (e.g., 14.5 per cent A1 2 3 ) is 
extracted from a liquid on the forsterite- 
albite join, the liquid will be enriched in 
silica and sodium disilicate, not nepheline: 

3Fo + 2Ab -> (Mg-Tsch + 5En) ss + 

Na 2 Si 2 5 + Si0 2 



Nepheline would appear along with so- 
dium disilicate only if excessive amounts 
of olivine were present. That is to 
say, the Tschermak's effect would be 
achieved only in picritic magmas. (The 
sodium disilicate would presumably com- 
bine with ferric iron in a magma to form 
the acmite molecule.) On the other hand, 
if the alumina content of enstatite is 
presumed to be present as the jadeite 
molecule, then nepheline components 
alone are enriched in the liquid: 

3Fo + 2Ab -> (Jd + 6En) ss + NaAlSi0 4 

Unfortunately, the jadeite molecule is not 
recognized in hypersthenes believed to 
come from depth; it is known in associ- 
ated clinopyroxenes from such environ- 
ments, however. There appear, therefore, 
to be other difficulties in the simple 
enstatite extraction method. 

The present study describes some 
serious experimental problems in attain- 
ing equilibrium at high pressures. We 
may begin to suspect not only the com- 
plete series of solid solutions presumed to 
exist along the join enstatite-jadeite but 
also the extensive field of enstatite on the 
liquidus of systems such as MgO-Al 2 3 - 
Si0 2 -H 2 at high pressure. 

The System Diopside-Forsterite- 
Enstatite at 20 Kilobars 

/. Kushiro 

Most primary basalt magmas originate 
in the upper mantle of the earth. There- 
fore, for an understanding of the origin of 
basalt magmas, it is desirable to know the 
liquid-solid equilibria at high pressures in 
the systems containing the components of 
the minerals in the upper mantle. Petro- 
logical studies indicate that forsterite-rich 
olivine, and diopside- and enstatite-rich 
pyroxenes, are the principal minerals in 
the Alpine-type peridotites and in the 
peridotite inclusions in basalts and kim- 
berlites. Since such olivines and pyroxenes 
are probably closely related to those 
existing in the upper mantle of the earth, 
the system containing diopside, forsterite, 
and enstatite is considered one of the 

fundamental systems. In the present 
experiments, the system diopside-forster- 
ite-enstatite was studied at 20 kb, 
corresponding to the pressure at the depth 
of 60 km. 

In the present experiments the solid- 
media high-pressure apparatus of the 
piston-cylinder type similar to that 
designed by Boyd and England (Year 
Books 57 and 60) was used, and crystal- 
lized glasses prepared by Kushiro and 
Schairer (Year Book 62, pp. 95-103) and 
by Boyd and Schairer (Year Book 61, pp. 
68-75) were the starting materials. 

The Join Diopside-F 'or sterile at 20 Kilobars 

Figure 26 shows the results of experi- 
ments along the join diopside-forsterite at 
20 kb near the liquidus temperatures. 
The piercing point exists at about 
Di 77 Fo 2 3 (weight per cent), which is about 
12 per cent more forsterite rich than that 
at 1 atm (DiggFon) shown in the lower 
part of the figure. The temperature of the 
piercing point is 1635° =b 10°C, about 
250° higher than that at 1 atm. It is 
noticed that the temperature difference 
between the melting point of diopside and 
the piercing point is still small and the 
liquidus of diopside very flat. For compo- 
sitions more diopside rich than Di 8 oFo 20 , 
the charges quenched from temperatures 
above 1645°C consist of large, fibrous 
quench clinopyroxene crystals, whereas 
those quenched from temperatures below 
1630°C consist of aggregates of subhedral 
or euhedral crystals of diopside or diop- 
side and forsterite, showing no evidence 
of melting. It is therefore indicated that 
the temperature difference between the 
liquidus and the solidus of diopside solid 
solution is less than 15°. The temperature 
maximum on the diopside solid solution 
liquidus found by Kushiro and Schairer 
(Year Book 62, pp. 95-103) at 1 atm was 
not demonstrable, if it exists, at 20 kb. 

The diopside and forsterite crystallizing 
from compositions on this join are not 
pure crystals of CaMgSi 2 6 and Mg 2 Si0 4 , 
respectively, but solid solutions whose 
compositions do not lie on this join. The 





Few + L 

Foss + Di< 

1300 - ,' 

CaMgSi 2 6 90 







Mg 2 Si0 4 

Weight % CaMgSi 2 6 

Fig. 26. Equilibrium diagram of the system diopside (CaMgSi 2 6 )-forsterite (Mg 2 Si0 4 ) at 20 kb 
pressure. Di 88 , diopside solid solution; Fo ss , forsterite solid solution; L, liquid. Diagram at 1 atm is 
from Kushiro and Schairer (Year Book 62, p. 96). 

X-ray powder patterns of the diopside 
and forsterite solid solutions are dis- 
tinctly different from those of pure 
diopside and forsterite, respectively. The 
A20[20(311) - 20(310)] values of the diop- 
side solid solutions are shown in table 3. 
The table shows that for most of the 
solid solutions the A20 is considerably 
smaller than that of pure diopside, and 
also that the A20 is smaller for the diop- 
side solid solutions crystallized from the 
mixtures more enriched in forsterite. The 
smaller A20 values are most probably 
attributed to the solid solution involving 
enstatite molecule as at 1 atm ( Year Book 
62, pp. 95-97). The A20 of the diopside 
solid solution decreases with increasing 
MgSi0 3 content (Year Book 62, p. 99), 
indicating that the diopside solid solu- 

TABLE 3. A20[20(311) - 29(310)] Values of 

Diopside Solid Solutions Crystallized from the 

Join Diopside-Forsterite at 20 Kilobars 

(CuK« radiation) 

Composition of 



A20 Value, 



of Run, °C 






0.614 ±0.006 




0.583 ±0.005 




0.590 ±0.013 




0.542 ±0.013 




0.543 ±0.010 




0.519 ±0.006 




0.518 ±0.011 




0.485 ±0.013 




0.270 ±0.030 

* Mechanical mixture. 



tions crystallized from the mixtures more 
enriched in forsterite would have compo- 
sitions more MgSi0 3 -rich. 

The d values of (021), (130), and (112) 
planes of the forsterite crystallizing from 
compositions on this join are larger than 
those of pure forsterite, indicating that 
Ca replaces a part of Mg of forsterite and 
forms the larnite (Ca 2 Si0 4 ) molecule. 
This evidence indicates that the join 
diopside-forsterite is not binary at 20 kb 
and that there must be a region of three- 
phase assemblage Fo ss + Di S3 + L just 
below the regions Di s3 + L and Fo ss + L. 
Consequently the join diopside-forsterite 
is not the thermal barrier, and the liquids 
initially having a small amount of ensta- 
tite component can change their compo- 
sitions toward merwinite (Ca3MgSi 2 8 ) 
across it. This fact is significant for the 
fractional crystallization of magmas of 
basalt composition at high pressures. 

That diopside probably dissolves a 
small amount of forsterite in solid solution 
at temperatures near the beginning of 
melting was ascertained by the same 
procedures as those used at 1 atm (Year 
Book 62, pp. 95-97). A mechanical mix- 
ture consisting of 95 weight per cent pure 
diopside and 5 per cent forsterite was 
heated at 1600°C for 10 minutes. After it 
was heated the forsterite reflections in the 
X-ray powder pattern disappeared. The 
mixture Di 92 Fo 8 , heated at 1600°C, shows 
forsterite in its X-ray powder pattern. 
The maximum content of forsterite may 
be less than 5 weight per cent near 1600°C 
at 20 kb and would be further reduced 
with decreasing temperature. 

The Join Wollastonite-Enstatite 
at 20 Kilobars 

The phase equilibria along the join 
wollastonite (CaSi0 3 )-enstatite (MgSi0 3 ) 
have been studied at 20 kb to aid in 
determining the phase relations in the 
system diopside-forsterite-enstatite. The 
results are shown in figure 27. 

The join wollastonite-diopside. The join 
wollastonite-diopside includes the stable 
phases wollastonite, pseudowollastonite, 

and diopside solid solutions near the 
liquidus and the solidus temperatures at 
20 kb. Pseudowollastonite is stable at 
temperatures between 1550° ± 15°C and 
its melting point, 1620° ± 15°C. It prob- 
ably forms a solid solution with diopside, 
although the content of diopside may be 
less than 5 weight per cent. Wollastonite 
is stable below 1550° ± 15°C and is in 
solid solution with diopside. The maxi- 
mum content of diopside in wollastonite 
is about 20 weight per cent. Diopside is 
stable at any temperature below its 
melting point at 20 kb and is in solid 
solution with wollastonite, although the 
content of wollastonite is small. The mix- 
ture Di 98 Wo 2 (weight per cent) does not 
contain wollastonite at 1530°C, whereas 
the mixture Di 95 Wo 5 contains rare wollas- 
tonite at the same temperature. Addi- 
tional evidence for this solid solution is 
that the intensities of wollastonite relative 
to diopside in the X-ray powder patterns 
of the mixture Di 90 Woio were much re- 
duced after heating at 1530°C and were 
nearly the same as those of the unheated 
mixture Di 95 Wo 6 . This evidence indicates 
that the maximum content of wollastonite 
is between 2 and 5 weight per cent. 

The eutectic between wollastonite and 
diopside is located at about Di 38 Wo 62 at 
1535° ± 15°C. The position of the 
eutectic at 20 kb is about 22 weight per 
cent more wollastonite rich than that at 
1 atm, and the temperature is about 170° 

The join diopside-enstatite. In the join 
diopside-enstatite, forsterite does not 
appear at any temperature, and all phases 
have compositions on this join at 20 kb. 
Therefore the join is binary at this 

The liquidus of diopside is very flat. 
For compositions between diopside and 
Di 6 oEn 4 o (CaSi0 3 32.2 per cent), the 
charges quenched from the temperatures 
above 1650°C consist of large, fibrous 
quench pyroxene crystals showing wavy 
extinction under the microscope, whereas 
those quenched from the temperatures 
below 1635°C consist of aggregates of 





£ 1500 






En ss +Di ss 


20 30 40 CaMgSi 2 6 60 

Weight % CaSi0 3 



90 CaSi0 3 

Fig. 27. Equilibrium diagram of the system enstatite (MgSi.0 3 )-wollastonite (CaSi0 3 ) at 20 kb 
pressure. En ss , enstatite solid solution; Pwo ss , pseudowollastonite solid solution; Wo ss , wollastonite 
solid solution; other abbreviations as in figure 26. 

small subhedral clinopyroxene crystals 
showing no evidence of melting. The 
temperature difference between liquidus 
and solidus of diopside is, therefore, less 
than 15°. 

The liquidus of enstatite was deter- 
mined for compositions more diopside rich 
than Di 4 oEn 60 (CaSi0 3 21.5 per cent). It 
has a steep slope compared with the liquid- 
us of diopside. The charges quenched from 
the temperatures above the enstatite 
liquidus consist of large quench pyroxene 
crystals. Below the liquidus, but above 
the solidus, charges contain orthoensta- 
tite crystals and usually quench crystals. 
The orthoenstatite crystals are sometimes 
rimmed by an undetermined fibrous 
clinopyroxene overgrowth. 

A peritectic point exists at about 
Di 60 En 4 o (CaSi0 3 32.2 per cent) at 1650° 
± 15°C, where enstatite solid solution 
reacts with liquid to produce diopside 
solid solution. The liquidus of diopside 
solid solution is believed to have a mini- 
mum between the peritectic point and 
diopside, but the relationship is not 

demonstrable within the error of the 

The diopside solid solution extends at 
least up to the composition Di 55 En 4 5 
(CaSi0 3 30 per cent) near the solidus 
temperatures. The mixture of very fine- 
grained diopside and enstatite solid solu- 
tions were homogenized into a single 
diopside solid solution up to the dashed 
line shown in the subsolidus region 
between diopside and enstatite of figure 
27. A mechanical mixture consisting of 20 
weight per cent pure enstatite and 80 per 
cent diopside solid solution of composition 
Di 7 5En 2 5 was also homogenized into a 
diopside solid solution at 1540°C, on the 
basis of the X-ray powder patterns. In the 
compositions more enstatite rich than the 
dashed line but less than Di 40 En 6 o (CaSi0 3 
21.5 per cent), two pyroxenes appear to 
be homogenized into a clinopyroxene. 
However, the X-ray powder patterns of 
the clinopyroxenes show a (231) plane 
reflection that is derived not from the 
diopside structure, C2i/c, but from the 
clinoenstatite structure, P2i/c. This evi- 



dence suggests that the clinopyroxenes 
more enstatite rich than the dashed line 
in the subsolidus region of figure 27 may 
be two phases or may have a different 
structure from diopside. Therefore it is 
not certain whether the dashed line 
represents the diopside-rich limb of the 
solvus or not. 

The diopside-rich limb of the solvus at 
high pressures may be located closer to 
diopside than that at 1 atm, as discussed 
below. In the reaction (Di,En) ss = Di S3 
+ En ss , if the volume of Di ss + En ss is 
smaller than that of the solid solution 
(Di,En) s8 , the reaction would proceed 
toward the right with increasing pressure, 
since the entropy of the smaller-volume 
phase is usually smaller than that of the 
larger-volume phase. The diopside-rich 
limb of the solvus shifts, therefore, toward 
diopside, and the enstatite-rich limb 
toward enstatite, with increasing pres- 

The unit cell volume of the diopside 
solid solution (Di 6 oEn 4 o) S s (by weight) 
measured by Clark, Schairer, and de 
Neufville (Year Book 61, p. 64) is 435 ± 
1 A 3 , whereas the total unit cell volume 
of 60 weight per cent pure diopside and 
40 per cent pure enstatite is calculated as 
425.4 A 3 on the basis of the unit cell 
volume of diopside given by Clark, 
Schairer, and de Neufville (Year Book 61, 
p. 62) and that of enstatite given by 
Morimoto, Appleman, and Evans (Year 
Book 58, p. 197). It is clear that the total 
unit cell volume of 60 per cent pure 
diopside and 40 per cent pure enstatite is 
smaller than that of the solid solution 
(Di 6 oEn 4 o)ss, and therefore the molar 
volume is also smaller. Consequently it is 
suggested that the reaction (Di 6 oEn 4 o) S s = 
0.6Di + 0.4En proceeds toward the right; 
that is, the diopside-rich limb of the 
solvus shifts toward diopside with in- 
creasing pressure. The total molar volume 
of Di S3 + En s3 is also estimated to be 
smaller than that of the diopside solid 
solution (Di,En) ss of the same composi- 
tion, and the reaction (Di,En) ss = Di as + 
En ss proceeds toward the right with 

increasing pressure. The dashed line in 
the subsolidus region of figure 27 repre- 
sents the outer limit of the diopside-rich 
limb of the solvus at 20 kb, since two 
separate pyroxenes were homogenized 
into a diopside solid solution of the 
diopside-rich side of the dashed line. 
Therefore, the actual solvus at this pres- 
sure would be located between the dashed 
line and the solvus at 1 atm which has 
been determined by Boyd and Schairer 
(Year Book 61, p. 70) and Kushiro and 
Schairer (Year Book 62, p. 98). 

Enstatite solid solution contains 7.5 to 
10 per cent diopside molecule near the 
solidus temperatures. It is almost the 
same as that determined at 30 kb by 
B. T. C. Davis (Year Book 62, pp. 

The System Diopside-Forsterite-Enstatite 
at 20 Kilobars 

The liquidus diagram of the system 
diopside-forsterite-enstatite at 20 kb, 
shown in figure 28, is based on the results 
of the present experiments in this ternary 
system and those of the joins diopside- 
forsterite and diopside-enstatite at 20 kb. 
In the same figure the results of the 
system diopside-forsterite-silica at 1 atm 
(Bowen, 1914; Schairer and Yoder, Year 
Book 61, pp. 75-82; and Kushiro and 
Schairer, Year Book 62, pp. 95-103) are 
shown for comparison. 

The FOss-Di ss boundary and the Fo ss - 
En S3 boundary shift toward forsterite 
with increasing pressure, and consequent- 
ly the liquidus surface of forsterite solid 
solution at 20 kb is considerably smaller 
than that at 1 atm. Forsterite crystal- 
lizing near the liquidus and the solidus 
temperatures in this system is a solid 
solution probably containing a small 
amount of Ca as inferred from the large d 
values as compared with those of pure 

On the Foss-pyroxene boundary curves 
there is an invariant point at 1640° ± 
10°C (P of fig. 28). It is most likely that 
it is a reaction point where the reaction 
En ss + L = Di sa + Fo ss takes place, and 



CaMgSi 2 6 


Mg 2 Si0 4 


Weight per cent 

Fig. 28. Preliminary diagram of the system diopside (CaMgSi 2 06)-forsterite (Mg 2 Si0 4 )-enstatite 
(MgSi0 3 ) at 20 kb pressure. Abbreviations as in figures 26 and 27. Diagram of the system diopside- 
forsterite-silica at 1 atm is from Bowen (1914), Schairer and Yoder (Year Book 61, p. 76), and Kushiro 
and Schairer (Year Book 62, p. 100). Italic letters are for diagram at 1 atm. Cr, cristobalite; Pr 8S , 
protoenstatite solid solution; Tr, tridymite. 

that the temperature of the Fo S s-Di ss 
boundary drops continuously from the 
reaction point to another invariant point, 
Fo ss + Di ss + Merw (merwinite, 
Ca 3 MgSi 2 8 ) + L in the system Fo-Di- 
Merw across the Di-Fo join. 

There is, however, another possibility, 
that the invariant point P is a ternary 
eutectic Fo 3S + Di ss + En sa + L and that 
the Fo ss -Di ss boundary curve has a 
thermal maximum between the invariant 
point P and the Di-Fo join. This possi- 
bility is considered because the Fo ss -Di 8S 

boundary curve has a temperature maxi- 
mum between the joins Di-En and Di-Fo 
at 1 atm (Year Book 62, pp. 100-101). In 
the present experiments it was not 
possible to determine which possibility is 
correct, because the temperature differ- 
ence between the invariant point P and 
the piercing point on the Di-Fo join is 
very small and within the error of the 

As was mentioned before, the Di-Fo 
join is not binary and consequently it is 
not a thermal barrier (this report, p. 103). 



Therefore, even if a thermal maximum is 
present on the Fo ss -Di s3 boundary be- 
tween the invariant point P and the 
Di-Fo join, the liquids containing a small 
amount of enstatite component change 
their compositions toward the invariant 
point Fo sa + Di ss + Merw + L across the 
Di-Fo join, which is a critical join of silica 
undersaturation in the system CaO-MgO- 
Si0 2 . 

At 1 atm, the reaction point (A of fig. 
28) is in the silica-oversaturated part of 
the system Di-Fo-silica. The reaction at 
A is Fo ss -\- L = Di S3 + Pr ss (protoensta- 
tite solid solution) {Year Book 62, pp. 
100-101). The differences in position and 
nature of the invariant points between 1 
atm and 20 kb are significant for under- 
standing the melting of peridotites and 
the fractional crystallization of basaltic 
magmas at different pressures. 

Most peridotite inclusions in basalts 
and kimberlites are rich in forsterite and 
enstatite relative to diopside (Ross, 
Foster, and Myers, 1954; O'Hara and 
Mercy, 1963), their compositions being 
plotted in an area close to the Fo-En join 
in the system Di-Fo-En when FeO, A1 2 3 , 
and other minor components are excepted. 
If the effects of these components on the 
phase equilibria in the system Di-Fo-En 
are small enough, the partial melting of 
these peridotites produces liquids over- 
saturated with silica (tholeiitic liquids) at 
1 atm or low pressures, since the invariant 
point A lies within the system Di-En- 
silica. At 20 kb, however, the partial 
melting of the same peridotite produces 
liquids undersaturated with silica (olivine 
tholeiitic liquids), since the invariant 
point P lies within the system Di-Fo-En. 
Thus the partial melting of the peridotites 
at different pressures would produce 
liquids significantly different in chemical 

If the fractional crystallization of 
liquids whose compositions are within the 
triangle Fo as -C-En (olivine tholeiitic liq- 
uids) takes place at 1 atm, the liquids 
change their compositions toward silica 
across the Di-En join, and the liquids 

oversaturated with silica are produced. 
If, on the other hand, the fractional 
crystallization of these liquids takes place 
at 20 kb, the liquids may change their 
compositions toward the Di-Fo join and 
further to the invariant point Di + 
Merw + Fo + L across the Di-Fo join, 
and liquids containing larnite (Ca 2 Si0 4 ) 
component (alkali basaltic liquids) may 
be produced, provided that the invariant 
point P is a reaction point. 

The liquids whose compositions are 
within the triangle Fo ss -C-Di ss can change 
their compositions toward akermanite 
across the Di-Fo join at 1 atm (Year Book 
62, p. 103). At 20 kb, and even if the 
invariant point is ternary eutectic, the 
liquids containing a small amount of 
enstatite component also change their 
compositions across the Di-Fo join and 
liquids containing larnite component are 

The origin of tholeiite, olivine tholeiite, 
and alkali basalts is one of the important 
problems in igneous petrology. Although 
many contributions have been made, 
many unsolved problems still remain. 
The present experiments suggest the 
following possibilities. 

1. Magmas of olivine tholeiite compo- 
sition can be produced by the partial 
melting of peridotites consisting mostly 
of forsterite, enstatite, and diopside at 
pressures near or higher than about 10 
kb. 7 

2. Magmas of tholeiite composition can 
be produced by the fractional crystalliza- 
tion of a magma of olivine tholeiite 
composition at pressures at least less than 
about 10 kb. 7 Magmas of tholeiite compo- 
sition can also be produced by the partial 
melting of peridotites at pressures at least 
less than about 10 kb. 7 However, it is not 
likely that magmas of tholeiite compo- 

7 Ten kilobars is considered because the pres- 
sure at which the join diopside-enstatite becomes 
a thermal barrier is equal or close to the pressure 
at which pure enstatite begins to melt congru- 
ently, that is, at least as low as 15 kb and 
probably as low as 6 kb (Boyd and England, 
Year Booh 60, p. 115). 



sition can be produced at pressures near 
or higher than about 10 kb 7 either by the 
melting of peridotites consisting of for- 
sterite, diopside, and enstatite or by the 
fractional crystallization of a magma of 
olivine tholeiite composition. 

3. Magmas of alkali basalt composition 
may be produced by the fractional 
crystallization of a magma of olivine 
tholeiite composition at pressures prob- 
ably higher than about 10 kb, 7 provided 
that the invariant point in the system 
Di-Fo-silica is a reaction point. If, on the 
other hand, the invariant point is eutectic- 
like, it would not be possible to derive 
magmas of alkali basalt composition at 
high pressures by the fractional crystal- 
lization of a magma of olivine tholeiite 
composition except for compositions con- 
taining a small amount of enstatite 
component or by the melting of perido- 
tites consisting of forsterite, diopside, and 

Experimental Studies on the 

Basalt-Eclogite Transformation 

/. Kushiro and H. S. Yoder, Jr. 

Transformation of basalt or gabbro to 
eclogite has been considered to be one of 
the important problems not only in 
petrology but also in geophysics, since the 
continental Mohorovicic discontinuity, as 
discussed by many authors, may be 
attributed to the basalt-eclogite trans- 
formation. It is, therefore, necessary to 
determine the physical conditions and the 
nature of the basalt-eclogite transforma- 
tion. Although Yoder and Tilley (1962) 
estimated the physical conditions of the 
transformation on the basis of experi- 
ments on natural rocks, the conditions 
and nature of the transformation are still 
not known in detail. One of the methods 
in outlining the physical conditions of the 
basalt-eclogite transformation is to deter- 
mine the univariant curves of the simple 
reactions by which components of basalt 
transform into those of eclogite. Several 
univariant curves bearing on the basalt- 
eclogite transformation have been deter- 
mined. Robertson, Birch, and MacDonald 

(1957) determined the univariant curve 
for the formation of jadeite from nephe- 
line and albite, and Birch and LeComte 
(1960) determined the univariant curve 
for the breakdown of albite into jadeite 
and quartz. Yoder (1955), Boyd and 
England {Year Book 58, pp. 83-87), and 
Yoder and Chinner (Year Book 59, pp. 
78-84) determined the stability fields of 
almandite and pyrope. There remain 
several important reactions whose uni- 
variant curves would be useful in clarify- 
ing the physical conditions of the basalt- 
eclogite transformation, namely, 

Albite + forsterite = (jadeite, 
2 enstatite) ss or jadeite + 

2 enstatite (1) 
Anorthite -f forsterite = garnet (2) 

Anorthite + 2 enstatite 

= garnet + quartz (3) 

3 Anorthite = grossularite -+- 

2 kyanite + quartz (4) 

4 Anorthite -f- diopside = 2 gar- 

net + 2 kyanite + 2 quartz (5) 

Some preliminary data are available on 
reactions 1 and 4 (Yoder and Tilley, 
1962) and on the composition anorthite + 
forsterite at high pressures (Yoder and 
Chinner, Year Book 59, pp. 78-84; 
O'Hara, Year Book 62, pp. 116-118). 

In the present experiments, reactions 2 
and 3 were studied. Reaction 2 indicates 
the formation of garnet (pyrope 66.7, 
grossularite 33.3, molecular per cent) from 
anorthite and forsterite and one of the 
critical reactions for the transformation 
of silica-undersaturated basalts to eclo- 
gite. Reaction 3 shows the formation of 
the same garnet in silica-saturated con- 
ditions and one of the critical reactions 
for the transformation of silica-saturated 
basalts to quartz eclogite. In the course 
of the present studies it was found that 
the reaction for the formation of garnet 
from anorthite and forsterite or enstatite 
is not carried out directly as expressed by 
reactions 2 and 3 but is complicated 
because of the intermediate appearance 
of pyroxene-rich assemblages. Although 
the experiments have not been completed, 



the results so far obtained are significant 
for an understanding of the basalt- 
eclogite transformation. 

Pressure-Temperature Plane for 
Anorthite + Forsterite Composition 

The pressure-temperature diagram for 
anorthite + forsterite composition is 
shown in figure 29. A glass and a crystal- 
lized glass of the composition anorthite 
50, forsterite 50, molecular per cent (66.4 
and 33.6 weight per cent, respectively) 
were prepared as starting materials. They 
are isochemical with a garnet of the 
composition pyrope (Py) 66.7, grossularite 
(Gr) 33.3, molecular per cent. The gas- 
media high-pressure apparatus designed 
by Yoder (1950) was used for the runs at 
pressures lower than 10 kb, and the 
solid-media high-pressure apparatus simi- 

lar to that designed by Boyd and England 
(Year Books 57 and 60) was used for runs 
higher than 10 kb. As shown in the figure 
the anorthite + forsterite assemblage is 
stable up to curve A. On the higher 
pressure side of this curve forsterite dis- 
appears, whereas pyroxenes and spinel 
appear and the assemblage is anorthite + 
spinel + pyroxenes. The pyroxenes are 
both clinopyroxene and orthopyroxene 
solid solutions. Some Tschermak's mole- 
cules (CaAl 2 Si0 6 and MgAl 2 Si0 6 ) may 
enter into each pyroxene in solid solution. 
The reaction may be presented, therefore, 
by the following: 

2CaAl 2 Si 2 8 + 2Mg 2 Si0 4 

Anorthite Forsterite 

= CaMgSi 2 6 -nCaAl 2 Si0 6 + 

Aluminous diopside 

— ? 

Gar + (Cpx ss ) 


15 20 

Pressure , kb 



Fig. 29. Pressure-temperature plane for forsterite (Mg 2 Si0 4 ) + anorthite (CaAl 2 Si 2 8 ) composi- 
tion. An, anorthite; Cpx 8S , clinopyroxene solid solution; Fo, forsterite; Gar, garnet; L, liquid; Opx ss , 
orthopyroxene solid solution; Sp, spinel (MgAl 2 4 ). Cpx S8 in parentheses may be metastable. 



2MgSi0 8 -wMgAl 2 SiO« + 

Aluminous enstatite 

(1 - n)MgAl 2 4 + 


(1 - n)CaAl 2 Si 2 8 (n < 1) 


It is to be emphasized that the above 
reaction takes place at relatively low 
pressures; that is, it takes place at about 
9 kb at 1300°C and at about 8 kb at 

The assemblage anorthite + pyroxenes 
+ spinel is stable up to curve B. Between 
curves A and B the compositions of 
pyroxenes and the relative amounts of 
pyroxenes to anorthite and spinel would 
change with pressure. In the reaction 

CaAl 2 Si 2 8 + MgAl 2 4 

CaAl 2 Si0 6 + MgAl 2 Si0 6 



the volume of the right-hand assemblage 
is much smaller (>10.5%) than that of 
the left-hand assemblage. (In calculating 
the volume change, the molar volume of 
Ca-Tschermak's molecule as estimated by 
Kushiro [1962] and the molar volume of 
clinoenstatite, which must be larger than 
that of Mg-Tschermak's molecule, were 
used.) Therefore the reaction would pro- 
ceed toward the right with increasing 
pressure at constant temperature. Both 
Ca- and Mg-Tschermak's molecules 
formed by this reaction are included in 
solid solution with diopside and enstatite, 
which are also mutually soluble. Conse- 
quently the contents of Tschermak's 
molecules in pyroxenes increase, whereas 
amounts of anorthite and spinel decrease, 
with increase of pressure. The X-ray 
powder patterns show that the relative 
intensities of anorthite reflections are 
strong for the runs close to curve A, 
whereas they become weak close to curve 
B, indicating that the pyroxenes become 
more aluminous with increasing pressure. 
On the higher pressure side of curve B, 
garnet appears and the assemblage is 
garnet + clinopyroxene, although the 
amount of garnet is not large in most runs. 

However, the amount of garnet relative 
to clinopyroxene increases with increasing 
duration of run as inferred from the 
evidence that the relative intensities of 
the garnet reflections in the X-ray powder 
pattern for a 1-hour run were considerably 
stronger than those for a 20-minute run 
at 1450°C at 25 kb. Observations under 
the microscope show that garnet crystals 
are small after a 20-minute run, whereas 
they are relatively large and include many 
small clinopyroxene inclusions after a 
1-hour run. Growth of garnet at the 
expense of clinopyroxene is inferred. This 
evidence suggests that the clinopyroxene 
coexisting with garnet may be metastable, 
and the garnet of composition Pv66.7Gr 33 . 3 
may be the only stable phase on the 
higher pressure side of curve B. Then the 
reaction on the univariant curve B is as 
follows : 

CaMgSi 2 6 -nCaAl 2 Si0 6 + 

Aluminous diopside 

2MgSi0 3 -nMgAl 2 Si0 6 + 

Aluminous enstatite 

(1 - n)CaAl 2 Si 2 8 + 

(1 - n)MgAl 2 4 

= 2CaMg 2 Al 2 Si 3 12 (n < 1) 

Garnet (Py66.7Gr33.3> 

However, compositions of garnet coexist- 
ing with these clinopyroxenes are more 
pyrope rich than Py66.7Gr 33 .3. The cell 
edges of garnets crystallized at 21 kb and 
1450°C, 25 kb and 1450°C, and 28 kb and 
1480°C are 11.543 ± 0.005, 11.554 ± 
0.005, and 11.556 ± 0.005 A, respectively, 
and the compositions are estimated as 
Py 78±2 Gr 22 , Py 7 4±2Gr 2 6, and Py 7 4±2Gr 2 6, re- 
spectively, on the basis of the cell edge vs. 
composition curve by Chinner, Boyd, and 
England {Year Book 59, p. 77). Therefore 
the clinopyroxenes coexisting with these 
garnets must have compositions more 
calcium rich than Di 5 oMg-Tsch 5 o. These 
clinopyroxenes may be stable at pressures 
lower than the pressure at which the 
garnet join begins to be a complete solid 
solution as suggested by Yoder and Tilley 
(1962), and the garnet + clinopyroxene 
assemblage may be stable on the higher 



pressure side of curve B. In the present 
experiments it was not determined 
whether clinopyroxene is stable or meta- 

In the melting interval, spinel appears 
over a wide pressure range. At 1 atm the 
mixture anorthite 50, forsterite 50, molec- 
ular per cent begins to melt at 1320° ± 5° 
to produce Sp (spinel) + Fo (forsterite) 
+ L, which is replaced by Sp + L at 
1375°C, and spinel disappears at about 
1490°C (Andersen, 1915; Osborn and 
Tait, 1952; Chinner and Schairer, 1962). 
The same sequence is obtained probably 
up to a pressure between 12 and 16 kb. 
Higher than this pressure, forsterite does 
not appear in the melting interval but 
clinopyroxene solid solution (Cpx ss ) ap- 
pears, and the Sp + Cpx ss + L region is 
present below the temperature interval of 
Sp + L. The Sp + Cpx ss + L region 
continues to exist probably up to about 
30 kb. Below the temperature of Sp + 
Cpx ss + L region and lower than 21 kb, 
there may be a four-phase region Sp -f- 
Cpx ss + Opx ss + L or Sp + Cpx ss + An + 
L, which, however, was not observed in the 
present experiments. Higher than about 
21 kb there may be a region garnet -f L 
+ Cpx ss , which also was not observed. 

An important finding of the present 
experiments is that there exists a con- 
siderably wide stability field of a pyrox- 
ene-rich assemblage between the anor- 
thite + forsterite assemblage and garnet 
or the garnet + clinopyroxene assem- 
blage. In the stability field of the py- 
roxene-rich assemblage, a spinel + alumi- 
nous clinopyroxene + aluminous ortho- 
pyroxene assemblage would be formed 
when the forsterite to anorthite ratio is 2, 
and a spinel + forsterite + aluminous 
clinopyroxene + aluminous orthopyrox- 
ene assemblage would be formed when 
the ratio is larger than 2. This last is a 
typical assemblage of the peridotite 
inclusions in basalts as described by Ross, 
Foster, and Myers (1954). Recently, 
Green (1964) has described a high- 
temperature peridotite intrusion in the 
Lizard area, in which a primary peridotite 

consisting of olivine, aluminous diopside, 
enstatite, and spinel has recrystallized 
into a plagioclase-bearing peridotite con- 
sisting of olivine, plagioclase, diopside, 
enstatite, and chromite in the marginal 
zone of the intrusion. Green considered 
that the primary peridotite crystallized 
at higher pressures and the recrys- 
tallization would have taken place at 
lower pressures, and that the reaction 
of the recrystallization is aluminous 
diopside + aluminous enstatite — ■» anor- 
thite + forsterite + diopside + enstatite. 
The reaction is essentially the same as 
that on curve A, figure 29, and his 
deductions are consistent with the present 

The extensions of curves A and B 
would intersect the probable geothermal 
gradient curve between 5 and 6 kb and 
between 9 and 11 kb, respectively, sug- 
gesting that the pyroxene-rich assemblage 
is stable between about 18 and 33 km, 
and that shallower than about 18 km the 
anorthite + forsterite assemblage is 
stable, whereas deeper than about 33 km 
garnet becomes stable. Although the 
effects of albite, fayalite, and ferrosilite 
components on curves A and B are not 
known, it may be suggested that olivine 
gabbro transforms to plagioclase and 
spinel-bearing pyroxenite or olivine and 
spinel-bearing pyroxenite in the lower 
part of the continental crust. This possi- 
bility has already been suggested by 
Yoder and Tilley (1962). 

The present results also suggest that 
the density change of the basalt (or 
gabbro)-eclogite transformation is not 
sharp but must be gradual because of the 
intermediate appearance of a pyroxene- 
rich assemblage between basalt and 
eclogite and of the density change in the 
pyroxene-rich assemblage with pressure. 
Therefore a sharp change in seismic wave 
velocities would not be expected in the 
basalt (or gabbro) -eclogite transforma- 
tion, and it seems unlikely that the 
Mohorovicic discontinuity can be attrib- 
uted to the basalt (or gabbro) -eclogite 



Pressure-Temperature Plane for 
Anorihite + 2 Enstatite Composition 

The pressure-temperature diagram for 
a composition close to 33.3 anorthite, 66.7 
enstatite in molecular per cent (58.1 and 
41.9 weight per cent, respectively) or 
garnet (Pv66.7Gr 3 3.3) + quartz is shown in 
figure 30. A crystallized glass of the 
composition 59 anorthite, 41 enstatite, 
weight per cent, prepared by Hytonen 
and Schairer {Year Book 60, pp. 125-141) 
was used as the starting material. 

On the lower pressure side of curve C 
the anorthite + enstatite assemblage is 
stable, but on the higher pressure side of 
curve C anorthite disappears and the 
assemblage is aluminous diopside and 
enstatite solid solutions with or without 
quartz. These relations suggest that the 
following reaction begins with increasing 
pressure to proceed toward the right 
along curve C: 

2CaAl 2 Si 2 8 + 4MgSi0 3 

Anorthite Enstatite 

= CaMgSi 2 6 -CaAl 2 Si0 6 + 

Aluminous diopside 

2MgSi0 3 -MgAl 2 Si0 6 

Aluminous enstatite 

2Si0 2 


Some mutual solid solution of enstatite 
and diopside would be present. It is noted 
that the reaction between anorthite and 
enstatite takes place at higher pressures 
than that between anorthite and for- 
sterite which is shown as curve A, figure 
29. The difference in pressure between 
curves A and C is about 5 kb at tempera- 
tures between 1200° and 1300°C. 

On the higher pressure side of curve D, 
garnet begins to appear and the assem- 
blage is garnet + pyroxene with or 
without quartz. The pyroxene includes 
both clinopyroxene and orthopyroxene up 
to about 25 kb, above which, however, it 
includes only a clinopyroxene. Although 
garnet is small in amount, it forms large 
idiomorphic crystals and can be identified 
easily under the microscope. The amount 
increases with increasing duration of the 
runs, as indicated by experiment. The 
amount of garnet observed by both the 
microscope and the X-ray powder pattern 
was considerably larger in the 80-minute 



o. i300 


I 100 

1 1 

i i 



Sp + Cpx ss 

+ L 

/"""Gar + L 

/ Sp+L ^c^- 


^.w 0--x in 

QUI q 

AT ---<n ? X J2-; 


--lav^m mj^-a a k 

s m 

m m a 


""* ^-*=-H / 

-^--^jC bb 




a n^a c P x ss % 

Gar 4- Qz +(Px ss ) 

a/ + a 


4^o + An + L 

An + En 

/a o P x ss / 

4 "/ 

/ / 

6 D 


15 20 

Pressure , kb 



Fig. 30. Pressure-temperature plane for 2 enstatite (MgSi0 3 ) + anorthite (CaAl 2 Si 2 8 ) composi- 
tion. An, anorthite; Cpx 88 , clinopyroxene solid solution; En, enstatite; Gar, garnet; L, liquid; Opx 8S , 
orthopyroxene solid solution; Px B8 , pyroxene solid solution (< about 25 kb, Cpx 88 + Opx S8 ; > about 
25 kb, Cpx 89 ), which may be metastable; Qz, quartz. 



run at 24 kb at 1450°C than in the 30- 
minute run at the same pressure and 
temperature. This evidence suggests that 
pyroxenes coexisting with garnet may be 
metastable and garnet + quartz may be 
the stable assemblage on the higher pres- 
sure side of the curve D, although the 
possibility that pyroxenes coexisting with 
garnet are stable is still not discarded. It 
is noted that curve D is located close to 
curve B of figure 29, indicating that the 
formation of garnet would take place at 
almost the same but not identical con- 
ditions for both the silica-undersaturated 
and the silica-saturated environments. 
Consequently the stability field of the 
pyroxene-rich assemblage in the system 
anorthite + 2 enstatite is considerably 
smaller than that in the system anorthite 
+ forsterite shown in figure 29. 

In the melting interval forsterite 
appears on the liquidus at low pressures, 
whereas higher than about 10 kb but 
lower than about 25 kb spinel appears, 
and higher than about 25 kb garnet 
appears, on the liquidus. Since the liquids 
coexisting with the above minerals are 
oversaturated with silica in the present 
system at any pressure, the partial 
melting of any assemblage of composition 
anorthite + 2 enstatite can produce 
silica-saturated liquids at 1 atm. 

The results of the present experiments 
yield information on the stability fields of 
silica-saturated basalts and pyroxenites 
and the physical conditions of the forma- 
tion of quartz eclogites. By extrapolating 
curves C and D to the probable geother- 
mal gradient curve, it is suggested that 
silica-saturated basalts such as tholeiite 
and quartz gabbro are stable at depths 
shallower than about 33 km and that 
quartz-bearing pyroxenite is stable be- 
tween about 33 and 43 km. Of course these 
depth estimates are influenced by the 
effects of other components. However, it 
is to be emphasized that silica-saturated 
basalts are stable over a wider depth 
range in the continental crust than silica- 
undersaturated basalts, and that the 
depth range where quartz-bearing pyrox- 

enites are stable is smaller than the range 
where olivine and spinel- bearing pyrox- 
enites are stable. 

On the higher pressure side of curve D 
the garnet + quartz assemblage is stable 
and, therefore, quartz-hypersthene eclo- 
gites would be formed from silica- 
saturated pyroxenites and also from 
silica-undersaturated pyroxenites whose 
compositions are within the triangle 
garnet-enstatite-A in figure 31. Another 
reaction required for the formation of 
quartz eclogite is diopside + albite = 
omphacite + quartz. If the univariant 
curve for this reaction is on the higher 
pressure side of curve D, quartz eclogite 
cannot form near curve D. If anorthite is 
still stable on the higher pressure side of 
curve D, the granulitic assemblage garnet 
+ anorthite + quartz is obtained from 
anorthite-rich compositions under the 
physical conditions where olivine eclo- 
gites and quartz eclogites are stable as 
shown in figure 31. Probably, at some 
pressure on the higher pressure side of 
curve D, anorthite breaks down into 
grossularite + kyanite + quartz, and 
kyanite eclogites may be formed from 
rocks rich in the anorthite component. 


(Py 2 G 


Fig. 31. Phases and tie lines in the system 
anorthite (CaAl 2 Si 2 08)-forsterite (Mg 2 Si0 4 )- 
quartz (Si0 2 ). Point A shows 2 enstatite + 
anorthite or garnet + quartz composition. 
Py*Gri, pyrope (Mg 3 Al 2 Si 3 0i 2 ) 66.7, grossularite 
(Ca 3 Al 2 Si 3 0i 2 ) 33.3 molecular per cent. 



Therefore it is suggested that the physical 
conditions for the formation of eclogites 
of different chemical composition are 
considerably different. 

Pyroxene Fractionation in Mafic 

Magma at High Pressures and 

Its Bearing on Basalt Genesis 

C. E. Tilley and H. S. Yoder, Jr. 

The problem of the source rocks of 
basalts in the upper mantle has long been 
debated, and the relations of the two 
dominant basalt types — the tholeiite and 
alkali olivine basalt — have, likewise, been 
the subject of continuing discussion and 
controversy. In these discussions the two 
series as developed in the Hawaiian 
Islands have been central in much of the 
inquiry. Among the workers who have 
attempted a solution of the problem of 
the relationship of the two suites have 
been Powers (1935, 1955) and Macdonald 
(1949). Powers (1935) invoked the frac- 
tional removal of hypersthene 8 from a 
tholeiitic parent, but in his later contri- 
bution (1955) he suspected that other 
operating factors were superimposed on a 
fractional crystallization process. Mac- 
donald (1949) provided a set of extrac- 
tions from a "cafemic submagma" (nor- 
mative olivine content 28 per cent) to 
produce a succession of Hawaiian alkali 
types, the process involving the removal 
of 55-90 per cent crystals from his 
" cafemic submagma" parent. Kuno and 
co-workers (1957) suggested that partial 
melting of peridotite at great depths, 
where enstatite was presumed to melt 
congruently, would give rise to an 
"undersaturated or alkali olivine basalt 
magma." The process was not more 
closely defined, but the possibilities in the 
simplified nonfeldspathic system diopside- 
forsterite-enstatite at 20 kb have now 
been explored by Kushiro (this report, 
p. 106). These possibilities depend on the 
nature of the invariant point in the sys- 

8 Holmes and Harwood (1932) had earlier 
expressed the view that extraction of olivine and 
enstatite from a "primary peridotite magma" 
could give rise to melilite basalt type liquids. 

tern and of the boundary curve leading to 
the diopside-forsterite join. 

In 1960 Murata returned to the concept 
of pyroxene extraction to produce the 
alkali basalt series from a tholeiitic 
parent, utilizing as extract a pyroxene 
fraction crystallized in the groundmass of 
Hawaiian tholeiites; but the particular 
extraction method, though providing a 
residual liquid enriched in alkalies, also 
enhanced its normative hypersthene con- 
tent (Yoder and Tilley, 1962, pp. 415- 
416). The present writers invoked the 
fractionation of an eclogite derived 
through partial melting of a garnet 
peridotite to give alternatively a tholeiite 
or an alkali basalt type liquid. 

From high-pressure experimental runs 
on a glass approximating an olivine 
tholeiite (normative olivine 19.6 per cent), 
Green and Ringwood (1964) crystallized 
an orthopyroxene as the sole phase at 
1300°C and 20 kb, and with a picritic 
composition, again orthopyroxene, as the 
single phase at 20 kb and a temperature 
of 1325°C. 

Both the orthopyroxene and the suc- 
ceeding clinopyroxene in these runs, 
examined with the electron probe, were 
found to be notably aluminous. Extrac- 
tion of 20 and 25 per cent of the ortho- 
pyroxene from the olivine tholeiite glass 
gave residua calculated as nepheline 
normative. The assumed precipitation of 
such a large percentage of a single ortho- 
pyroxene phase without the onset of 
crystallization of a second clinopyroxene, 
which indeed follows in their runs, must 
be considered excessive. The residua were 
quite low in alkalies, consequent on the 
low alkalies in the parent glass. They are, 
however, also exceptionally high in lime 
(12.3, 12.7 per cent) for the average 
alkali olivine basalt, a result following on 
the extraction of hypersthene only. 

In the experiments now reported the 
crystallization of two Hawaiian tholeiite 
basalts — the 1840 picrite basalt of Nana- 
wale Bay, Hawaii, and the 1921 olivine 
tholeiite of Kilauea (table 4, analyses 1 
and 2) — have been studied at various 



TABLE 4. Chemical Analyses and Norms of 
Hawaiian Rocks, Including Average Analyses 








Si0 2 






A1 2 3 






Fe 2 3 






























Na 2 






K 2 






Ti0 2 






P 2 6 






Cr 2 3 



H 2 + 




H 2 0- 




100.21 100.30 100.04 100.00 100.00 





























































100.11 100.25 99.93 100.10 100.08 

1. Picrite basalt, lava of the lower vents of the 
1840 eruption of Kilauea, Nanawale Bay, Hawaii 
(Year Book 62, p. 79); F/(F + M) 0.37 4 (iron 
enrichment [FeO + Fe 2 3 ]/[MgO + FeO + 
Fe 2 3 ]). 

2. Olivine tholeiite, south edge of Kilauea 
Crater, eruption of 1921 (Yoder and Tilley, 1962, 
p. 363); F/(F + M) 0.51 4 . 

3. Hypersthene olivine basalt, east wall of 
Olokele Canyon, Kauai, Hawaiian Islands {Year 
Book 62, p. 79, analysis 4); F/(F + M) 0.39 9. 

4. Average analysis of the two Kilauean erup- 
tions of 1840, 1840 Pb (analysis 1) and the lava 
of the upper vents 1840 b i (Year Book 62, p. 79, 
analysis 2); F/(F + M) 0.45 8 . 

5. Average alkali olivine basalt of the Hawaiian 
Islands (7 analyses) corrected to 100 per cent 
with removal of H 2 (Kuno et al., 1957, table 10, 
analysis 2); F/(F + M) 0.55 4 . 

temperatures at 20 kb from the liquidus 
to temperatures where garnet and py- 
roxene are crystallizing together. For his 
assistance in conducting these high- 
pressure runs for the two rocks we are 
greatly indebted to Dr. Kushiro. 

The results of the runs on the 1921 
Kilauea lava and the 1840 picrite basalt 
are summarized in table 5, which also 
presents the results of the behavior of the 
1921 Kilauea lava at 1200°C under a 
pressure of 20 and 31.4 kb as previously 
reported (Yoder and Tilley, 1962, p. 497, 
table 50). 

For the 1921 lava, euhedral clino- 
pyroxene and rare garnet appeared in a 
fine aggregate of quench pyroxene set in 
glass at 1325°C, the liquidus lying be- 
tween this temperature and 1375°C, 
where the charge consisted wholly of 
quench pyroxene in glass. The quench 
pyroxene of these runs takes the form, 
not of the usual feathery growths in glass, 
but of blocky, almost equidimensional 
rectangular crystals with spiked out- 
growths at their coigns and less commonly 
along their edges. In addition, spikes in 
the glass frequently take the form of 
three-pronged rods. Variants of the 
blocky crystals have a pseudohexagonal 
appearance. Where primary clinopyrox- 
ene appears it is distinctly euhedral and 
shows no outgrowths. Garnet and clino- 
pyroxene prevail at lower temperatures. 
In none of the runs was either olivine or 
orthopyroxene observed. 9 

In the 1840 picrite basalt runs, the 
appearance of rare primary olivine at 
1525° is associated with the development 
of fibrous quench olivine in the glass. The 
liquidus is taken as about 1535°C, in 
comparison with a liquidus of 1435°C at 
1 atm. Olivine is joined at 1475°C by 
clear euhedral prisms of orthopyroxene 

9 At 1 atm Opx is observed in the 1921 lava in 
runs at 1100°C held for 2 weeks (cf. with run at 
1100°C for 1 hour, Yoder and Tilley, 1962, p. 
379). There is therefore the possibility that, with 
longer runs at these lower temperatures at 20 kb, 
Opx would crystallize late from the liquid along 
with Cpx. 



TABLE 5. Results of Runs Carried Out on the 1921 Kilauea Olivine 
Tholeiite and the 1840 Kilauea Picrite Basalt at 20 Kilobars 

T, °C t, hours 












(31.4 kb) 













1921 Olivine Tholeiite (table 4, analysis 2) 

Cpx (quench) glass 

Cpx, rare garnet, glass with quench Cpx 
Cpx, rare garnet, glass with quench Cpx 
Cpx, garnet, interstitial glass (n = 1.550) 
Cpx, garnet, little interstitial glass 

1840 Picrite Basalt (table 4, analysis 1) 

Rare 01 in glass with quench 01, n of glass 1.645 

01, Opx mantled by Cpx, rare Cpx, quench Cpx in glass 

01, Opx mantled by Cpx, Cpx, quench Cpx in glass 

01, reduced in amount, Opx with wide mantles of Cpx, Cpx abundant, 

little glass 
Rare garnet, some Opx, fine-grained Cpx, no 01, little glass, some 

brown spinellid (oxidation?) 
Rare garnet, fine-grained Cpx, traces of glass, no 01, no Opx observed, 

brown spinellid (oxidation?) 

mantled by clinopyroxene. Clinopyroxene 
is rare as independent crystals at this 
temperature, and the examples observed 
may be elongate fragments of the ortho- 
pyroxene sheaths. At 1450°C olivine is 
beginning to diminish in amount with 
mantled orthopyroxene and independent 
clinopyroxene as the prominent phases. 
Particularly noteworthy at 1400°C are 
the wide clinopyroxene mantles to the 
orthopyroxene prisms, which may be 
reduced to a thin midrib in the com- 
posite prisms. It is questionable whether 
these wide clinopyroxene rims can be 
interpreted as quench products. The 
orthopyroxene may enclose small rounded 
olivines appearing as if in the process of 
resorption. Glass is now quite subordi- 
nate. Garnet is rare at 1300°C, and olivine 
is no longer present. At 1200°C garnet 
remains rare and the charge consists 
almost wholly of fine-grained clinopy- 
roxene in which an orthopyroxene cannot 
be observed. A brown spinellid is inter- 
spersed in the pyroxene aggregates, both 
at this temperature and in the run at 
1300°C. Both these runs were longer, and 
the appearance of the spinellid may be 
the result of oxidation of the charge. 

The question of the stability of ortho- 
pyroxene in the higher temperature runs 
needs some consideration. The excessive 
enstatite that appeared at the presumed 
olivine liquidus of the albite-forsterite 
system at 9 kb (Yoder, this report, p. 99) 
was believed to be a metastable product. 
It grew rapidly in well formed rods 
clearly distinguishable from quenching 
products where olivine and liquid were 
expected. It is possible, therefore, that 
the orthopyroxene appearing in the runs 
described above as well formed crystals 
are metastable pyroxenes proxying for 
olivine. The thin-ribbed composite prisms 
may be viewed as metastable rods of 
hypersthene which formed rapidly and 
acted as nuclei for the slower-growing 
clinopyroxenes that sheathe them. The 
interpretation of the hypersthene in 
liquid as a metastable growth product 
constitutes a serious objection to the con- 
cept of orthopyroxene extraction as a 
normal fractionation process. The sta- 
bility problem cannot be resolved at this 
time, but even so the objection would not 
invalidate the separation of a hyper- 
sthenic clinopyroxene, or even a process 
in the magma chamber at depth which 



invoked separation of metastable prod- 

It should be noted that, in the crystal- 
lization of the 1840 picrite basalt, olivine 
and clinopyroxene must accompany the 
separation of an orthopyroxene, stable or 
metastable at the higher temperatures. 
The suppression of olivine in the runs on 
the 1921 Kilauea lava and its gradual 
diminution in the lower temperature runs 
on the 1840 picrite basalt are of great 
significance, for the substitute precipi- 
tation of pyroxene — whether of ortho or 
clino type — involves a concomitant de- 
pletion of the residual liquid in silica. The 
melting behavior of the Loch Duich 
eclogite (Yoder and Tilley, 1962, p. 493), 
a rock whose bulk chemical composition 
is tholeiitic, may be recalled. At 1 atm 
olivine appeared on the liquidus at 
1180°C, but at 10 kb the liquidus was at 
1225°C with clinopyroxene as the primary 
phase, and at 1150°C the charge consisted 
wholly of clinopyroxene. The behavior of 
this rock suggested that pyroxenites 
might exist at depth and constitute pos- 
sible source rock of basalts (Yoder and 
Tilley, 1962, p. 493). The present experi- 
ments strengthen this hypothesis. Thus 
at high pressures olivine reacting with 
anorthite of plagioclase gives place to 
clinopyroxene and orthopyroxene. In the 
equation usually formulated these phases 
are accompanied by spinel; see equation 1. 
Such a high- pressure reaction was pro- 
posed by Davidson (1943) for a group of 
metamorphosed ultrabasic rocks in South 
Harris, Outer Hebrides, and earlier had 
been applied to explain the pyroxene 
symplectites which intervene between 
olivine and basic plagioclase in trans- 
formed olivine gabbros and allivalites, 

though high pressures were not then 
inferred. That the assemblage is favored 
by high pressure is also supported by 
recent experimental studies (Yoder and 
Chinner, Year Book 59, pp. 78-81). These 
investigators obtained the assemblage 
over a range of compositions on the join 
grossularite-pyrope at 10,000 bars water 
pressure, and more recently it has been 
shown to be developed from a 1:1 
anorthite-forsterite crystallized glass in 
the 8-15 kb region (Kushiro and Yoder, 
this report, p. 109). 

In the present studies on the picrite 
basalt, spinel is not a significant phase, 
and we may write equation 2 (see below), 
the Tschermak silicate contributing the 
alumina to the new-formed as well as the 
associated clinopyroxene of the picrite 
basalt assemblages (Cpx + 0px) 83 or 
Cpx ss + Opx ss , at this high pressure. The 
albite part of the plagioclase may supply 
the additional silica (with a breakdown 
to a jadeite component) to silicate more 
olivine. Equation 1 can then be reformu- 
lated as shown in equation 3, below. 
That the pyroxenes produced at the high 
temperatures of these runs can be con- 
sidered as holding the Mg-Tschermak 
silicate is clear from the electron-probe 
analyses of the ortho- and clinopyroxenes 
produced in the runs on an olivine 
tholeiite type glass at 15 kb and 1300°C 
carried out by Green and Ringwood 
(1964). Study of their analytical data 
indicates that the two pyroxenes in ques- 
tion can be computed into the following 
components (weight per cent, approxi- 
mately) : at 15 kb, 1300°C, Opx = 
Di n Hy76Tsi 3 , Cpx = Di 52 Hy32Tsi6; at 
20 kb, 1300°C, Opx = Dii 2 Hy 69 Tsi 9 . 

The coexisting pyroxenes at 15 kb 

CaAl 2 Si 2 8 + 2Mg 2 Si0 4 -> CaMgSi 2 6 + 2MgSi0 3 + MgAl 2 4 (1) 

zCaMgSi 2 6 + CaAl 2 Si 2 8 + Mg 2 Si0 4 -> [(x + 1) CaMgSi 2 6 + MgAl 2 Si0 6 ] s8 (2) 

CaAl 2 Si 2 8 + NaAlSi 3 8 + 2Mg 2 Si0 4 + zCaMgSi 2 6 -» 

(s + l)CaMg Si 2 Q 6 + 2MgSiO» + MgAUSiOe + NaAlSi 2 6 (liq.) (3) 

CpXsa + Opxss or (Cpx + Opx)ss 



plotted in the pyroxene quadrilateral 
show a distinctive tie line comparable 
with those revealed for the coexisting 
pyroxene phases of nodules in alkali 
basalts (Brown, 1961). The potential 
jadeite from the breakdown of plagioclase 
(equation 3) would appear to be stored 
for the most part in the liquid from which 
these early pyroxenes are precipitated. 
When the norms of the two experimental 
treated rocks (1840 Pb , Kilauea 1921) are 
considered it will be noted that, although 
the picrite basalt has 30 per cent norma- 
tive olivine, the 1921 lava has only 9 per 
cent. The nonappearance of orthopy- 
roxene in the runs of the 1921 lava indi- 
cate that it is accommodated at the 
temperature in clinopyroxene as a sub- 
calcic augite. This proved no longer 
possible in the picrite basalt of unusually 
high normative olivine, and therefore 
high potential hypersthene, content. 

The reactions that ensue with falling 
temperature in both rocks lead to the 
formation of garnet as a constituent of an 
ultimate eclogite assemblage. Effectively, 
we have, therefore, a further transforma- 
tion of the right-hand side of equation 3 : 

CaMgSiA + MgAl 2 Si0 6 

-» CaMg 2 Al 2 Si 3 0i2 (4) 

the jadeite finally entering the eclogite 

The experimental work of O'Hara 
(Year Book 62, pp. 117-118) on the join 
diopside-pyrope at 30 kb is pertinent to 
the present studies. Enstatite is on the 
liquidus and presumed stable over a 
narrow range of compositions along this 
join around a composition Di 35 Pv65. Such 
a composition is essentially tholeiitic and 
can be expressed as An45En 32 Ol23 (weight 
per cent) or AniEn 2 01i (mole per cent). 
Such an assemblage at 30 kb develops 
enstatite on the liquidus at about 1640°C 
and at a reaction point (1625°C) reacts 
with liquid to produce garnet and clino- 
pyroxene. On either side of this compo- 
sition where clinopyroxene and garnet 
respectively join enstatite as crystallizing 
phases in the liquid the solidus is reached 

at 1625°C. The assemblages Gr ss + Cpx S3 
+ Opx ss can be considered the ultimate 
phases at the highest pressure (30 kb) 
over a wide range of compositions along 
the join. The same does not hold for 
simple eclogite compositions, however. 

In the light of the preceding discussion 
we may conclude that an anorthite- 
forsterite-enstatite assemblage of these 
compositions at 1 atm with increasing 
pressure goes through several stages 
either in the solid state or in the presence 
of a liquid phase according to the PT 
conditions. For composition near Di 35 Py 6 5 
on the diopside side, the reactions appear 
as follows, in the order of increasing 
pressure : 

2An + 4En + 201 

-> (a) Di + 6En + Sp + An 
-> (6) 2Di + 4En + 2Ts 
(= Di ss + En 9S ) 
-» (c) Gr ss + Di 8 s + En ss 

Of these, stage b corresponds to that 
deduced for the 1840 picrite basalt as set 
out in equation 2. 

In the light of the crystallization 
sequence, the reaction of orthopyroxene 
and liquid, and the known composition of 
the early pyroxenes in the runs carried 
out by Green and Ringwood (1964), we 
can attempt an extraction process utiliz- 
ing Hawaiian assemblages of known 
composition or their averages — all olivine 
tholeiitic, in conjunction with an average 
alkali olivine basalt composition, based on 
seven analyses (table 4, analyses 1-5 
with norms) . In the extraction process the 
limits of extraction have gone so far as to 
involve reduction of a particular oxide 
(K 2 ,P 2 5 ) in two examples to near zero. 
The results are set out in table 6, analyses 
1-3. The three extracts have in common 
a high content of normative hypersthene 
but variable normative olivine. The 
interpretation of these extracts as poten- 
tial pyroxene or pyroxenes is expressed in 
the last section of table 6, there calculated 
to a pyroxene formula (O = 6). 

The compositions as such, in their 
alumina content, are comparable with the 



TABLE 6. Results of Extraction from Tholeiitic Basalts and Averages of Table 4 
to Produce Alkali Olivine Basalt (Table 4, analysis 5) 








































Si0 2 

A1 2 3 

Fe 2 3 





Na 2 

K 2 

Ti0 2 

P 2 5 
















Compositions 1 to 3 Calculated to Pyroxene Formula 



Fe +2 









0.004 [ 






















1. Extract (60%) from Kauai hypersthene olivine basalt, table 4, analysis 
3; F/(F + M) 0.33 3 . / To give average alkali 

2. Extract (40%) from average Kilauea 1840 picrite basalt and 18401 olivine basalt of the 
basalt of upper vents, table 4, analysis 4; F/(F + M) 0.36 . ( Hawaiian Islands 

3. Extract (50%) from average Kilauea 1840 picrite basalt and 1921 \ (table 4, analysis 5) 
Kilauea olivine tholeiite, table 4, analyses 1 and 2; F/(F + M) 0.34 3 . 

pyroxenes of the experimental runs at 
high pressure to which reference has been 
made, and also with the pyroxenes of 
some of the nodules of alkali basalts 
(Ross, Foster, and Myers, 1954). As re- 
gards iron enrichment, F/(F + M), they 
range from 0.33 6 to 0.36 , and are thus less 
magnesian than the usual pyroxenes 
recorded from nodules; the experimental 
pyroxenes from the olivine tholeiite glass 
show a range from 0.17 4 to 0.36 , the 
latter, an Opx, at 20 kb and at a tempera- 
ture of 1200° to 1300°C. The remarkable 

circumstance that nodules carrying oli- 
vine, clinopyroxene, and orthopyroxene 
are almost wholly confined to alkali- type 
basalts has been commented on by a 
number of workers (Powers, 1955; Kuno, 
1959), and their significance has been 
interpreted in various ways. 

A study of a large number of nodules 
from Salt Lake Crater, Honolulu, and 
from basalts near Williams, Arizona, for 
access to which the writers are indebted 
to Dr. F. Chayes, demonstrates that 
assemblages carrying both ortho- and 



clinopyroxenes with olivine are dominant. 
The content of olivine is variable, and 
such that many of them are of pyroxenite 
type and should be so named, rather than 
grouped under the general designation 
peridotite nodules. At Salt Lake Crater 
such nodules are joined by others carrying 
garnet-eclogite types, which are them- 
selves often hypersthene bearing. Two 
theories of the origin of the nodules have 
been advanced: (1) The nodules are meta- 
morphosed fragments of the upper man- 
tle. This view is expressed by Ross, 
Foster, and Myers (1954) and supported 
by Wilshire and Binns (1961). (2) On the 
other hand, O'Hara and Mercy (1963, p. 
310) have described evidence that they 
are igneous accumulates and "may be a 
fundamental part of the process whereby 
alkali basalt and melilite basalt are de- 
rived by partial fusion of garnet perido- 
tites in the mantle." 

Theory 2 would make them residuals 
rather than accumulates. One feature of 
the nodules, apart from their variable 
olivine content, calls for comment. 
Whereas their pyroxenes are characteris- 
tically aluminous, these clinopyroxene 
phases show almost uniformly a high 
calcium content, notably higher than the 
values recorded for the experimentally 
crystallized subcalcic augite of the olivine 
tholeiite melt at high pressures. Whether 
this distinctive feature can be explained 
by adjustment of the clinopyroxene to 
higher lime values in the later history of 
the nodules — whether they have, in fact, 
been recrystallized at lower tempera- 
tures — is not clear. Certainly some nod- 
ules show features of metamorphism, and 
others develop at least considerable 
exsolution of an orthopyroxene phase 
from their clinopyroxenes; yet there are 
others having igneous textures whether as 
residuals or as accumulates with inter- 
precipitate material. Clearly, a closer 
study of such nodules in basalts, chem- 
ically and texturally, is much needed. 

The arguments presented are perhaps 
more in favor of the theory that these 

nodules are accumulates or residuals 
rather than fragments of the upper 
mantle unmodified. As accumulates they 
could be viewed as early crystallizations 
from a tholeiitic ultrabasic magma under 
high pressure yielding a residual liquid of 
alkali basalt facies and brought to the 
surface by eruption, and thus genetically 
related to it at depth. 

The time succession of Hawaiian 
magmas, a tholeiite phase followed by the 
much less voluminous "decadent" stage 
of alkali basalt, might reflect such a 
mechanism at depth. As residuals they 
could be interpreted as the unmelted 
refractory fraction of a parent under- 
going partial melting to give the alkali 
basalt type followed by tholeiite with 
more drastic partial melting, a succession 
known in other provinces. In either 
sequence the opportunity for partial or 
complete recrystallization of the nodules 
in the solid state might well be available. 

Pyroxenite stage in basalt genesis. The 
experimental behavior of the rocks under 
study at 20 kb and of the Loch Duich 
eclogite at 10 kb (and 30 kb) previously 
reported reveals that the phase sequence 
from the liquidus to the garnet pyroxene 
assemblages can be expressed as a suc- 
cession of transformations involving an 
intermediate pyroxenite stage, either at 
intermediate pressures or at higher tem- 
peratures at higher pressures. At pres- 
sures corresponding to upper mantle 
conditions, this pyroxenite stage is repre- 
sented by the early crystallization of 
pyroxenes having a reaction relation with 
liquid (O'Hara and Yoder, Year Book 62, 
p. 69). The extraction of these phases 
(orthopyroxene and clinopyroxene) from 
an olivine-rich tholeiitic liquid leads to 
the generation of an alkalic undersatu- 
rated residual liquid of the alkali olivine 
basalt type encountered in the Hawaiian 
magmatic succession. This mechanism 
may provide one likely source, both of 
alkali basalt and of tholeiite, according to 
the stage of crystallization or, conversely, 
the stage of partial melting. Such a stage 



precedes the onset of the garnet-ompha- 
cite association of the eclogites at higher 
pressures or lower temperatures. 

Isothermal Sections of 

Pyroxene Quadrilateral 

H. S. Yoder, Jr., C. E. Tilley, and J. F. Schairer 

The course of crystallization of the 
pyroxenes has held the attention of many 
workers. Most recently Muir (1951), 
Brown (1957), and Brown and Vincent 
(1963) have dealt with the Skaergaard 
trend; Carmichael (1960) examined the 
trend of some pyroxenes from British and 
Icelandic Tertiary acid glasses ; Wilkinson 
(1956) and Murray (1954) each defined 
trends for alkali basalt magmas; Hess 
(1960) presented data for the trends in 
the Stillwater and Bushveld complexes; 
and McDougall (1961) outlined the trend 
for the Red Hill intrusion in Tasmania. 
These trends are based on the analyses of 
pyroxenes which presumably grew in the 
presence of large amounts of plagioclase. 
In an effort to understand these various 
trends, the natural pyroxenes from some 
of the occurrences cited above as well as 
other pyroxenes and synthetic mixtures 
were heated at various temperatures to 
ascertain their phase relations, first in the 
absence of plagioclase. 

Some of the results of such thermal 
studies are portrayed in a series of 
isothermal projections onto the pyroxene 
plane of portions of the system CaO- 
MgO-FeO-Si0 2 . As examples of the types 
of projection obtained at this stage of the 
investigation, the 1350°, 1250°, 1150°, and 
1050°C sections are given in figures 32, 
35, 37, and 39, respectively. The working 
liquidus diagram was presented in last 
year's report as figure 16 (Year Book 62, 
p. 85), and the flow sheet was outlined in 
figure 18 (Year Book 62, p. 87). Although 
the liquidus diagram will require some 
minor modifications, it is adequate for 
discussing the general pyroxene relations, 
and it as well as the flow sheet should be 
consulted for the following discussion. 

The isothermal sections were chosen to 
portray the relations above the tempera- 
ture of point A in the flow sheet and 
liquidus diagram, a temperature some- 
what below A but above B, a temperature 
somewhat below C, and a temperature 
just above D. Work is now in progress to 
locate the temperature of D (flow sheet) 
and delineate the reaction Opx with 
liquid to give Cpx + 01 + Trd. This 
reaction marks the disappearance of 
pigeonite in some major layered intru- 
sions and is of vital import to the general 
course of basalt fractionation. Some 
additional diagrams are given which may 
aid the reader in appreciating the values 
of the isothermal projections, although 
some liberties were taken for clarity of 
presentation. The writers believe that the 
depicted relations can be readily visual- 
ized if attention is maintained on the 
composition of liquid in equilibrium with 
crystals. It may be on, above, or below 
the pyroxene plane of composition, de- 
pending on the crystals formed; that is, 
the liquid may be of a pyroxene compo- 
sition, enriched in silica, or enriched in 
olivine components. 

1350°C section. The four curves bound- 
ing the liquid (L) field mark the 1350°C 
isotherm (fig. 32). All major phases, Cpx, 
Opx, 01, and Trd, have a field of stability 
in the presence of liquid. The composition 
of liquids in equilibrium with only Opx or 
only Cpx lies in the pyroxene plane; 
olivine solid solutions are in equilibrium 
with liquids whose composition lies above 
the pyroxene plane; and tridymite is in 
equilibrium with liquids below the plane. 
A schematic section through the tetra- 
hedron in the region of the olivine and 
pyroxene composition planes for this 
temperature was constructed along the 
line a-a' (figs. 32 and 33). The section was 
chosen so that it would be approximately 
parallel to one of the family of three-phase 
triangles, Opx + 01 + L. Another sec- 
tion, through a portion of the MgO-FeO- 
Si0 2 face of the tetrahedron, portrays 
schematically the relations at the base of 



the pyroxene quadrilateral En-Fs (figs. 
32 and 34). 

With decreasing temperature the oli- 
vine field gradually diminishes in size for 
compositions in the pyroxene plane, and 
at a temperature near 1300°C olivine no 
longer appears in the assemblages. Point 
A (Year Book 62, p. 85, fig. 16), about 
1300°C, in the liquidus diagram is a 
piercing point and does not mark a 
special thermal event ; it is merely a point 
on the four-phase curve Opx + Cpx + 
01 + L. 

1250°C section. The four curves mark- 

ing where the all-liquid region surfaces 
pass through the pyroxene plane also 
limit the fields in which the major phases 
are in equilibrium with liquid : Opx + L, 
Cpx + L, Trd + L, and Wo + L, just off 
the pyroxene quadrilateral (fig. 35). A 
schematic section through a portion of 
the principal tetrahedron along the lines 
a-a'-a" is given in figures 35 and 36. A 
similar section through the Opx + L 
region would show that the L region near 
b (fig. 36) lies above the pyroxene plane, 
and the Trd + L and Opx + Trd + L 
fields are absent for compositions in the 


Wo + L 



Cpx + OI + LVgJ* 

Opx+Cpx+OI + L 

Weight per cent 

Fig. 32. Preliminary 1350°C diagram for compositions in the pyroxene plane of the CaO-MgO- 
FeO-Si0 2 system. Relations along the section a-a' are portrayed in figure 33. Cpx, clinopyroxene; 
Di, diopside; En, enstatite; Fs, ferrosilite; Hd, hedenbergite; L, liquid; 01, olivine; Opx, ortho- 
pyroxene; Pwo, pseudowollastonite; Trd, tridymite; Wo, wollastonite. 



pyroxene plane. Attention is called to the 
generation of the four-phase volume, 
01 + Opx + Trd + L. 

With further decreasing temperature, 
the all-liquid region drops below the plane 
of composition of the pyroxenes, and Trd 
becomes an important phase in most of 
the assemblages. 

1150°C section. Approximately half the 
pyroxene quadrilateral is crystalline at 
this temperature (fig. 37). The critical 
assemblage Cpx + Opx + L has ad- 
vanced toward more iron-rich composi- 
tions. The Cpx + Wo + Trd + L region 
now dominates the region near the 
hedenbergite composition, which is there 
represented by Wo + Trd + L. A 
schematic section, a-a'-a", is given in 
figures 37 and 38 to illustrate the relative 
position of the liquid field which is wholly 
below the pyroxene plane of compositions. 
In view of the fact that Cpx and Trd are 
in equilibrium, it is concluded that the 
temperature is well below the piercing 
point B in the liquidus diagram (Year 
Book 62, p. 85, fig. 16). As Wo and Trd 

are also in equilibrium, it is evident that 
the temperature of the section is below 
that of the piercing point C (Year Book 
62, p. 85, fig. 16). The point B is esti- 
mated to be about 1200°C; and C, 
approximately 1175°C. 

1050°C section. The principal change 
taking place with temperature decreasing 
from 1150°C is the generation of the 
Wo + 01 + Trd + L volume in the iron- 
rich portion of the quadrilateral (fig. 39). 
Tridymite is now present in all assem- 
blages in which liquid is present and in 
some of the all-crystalline fields. 

Of note is the fact that the closure of 
the Trd + L field in the pyroxene plane 
is approaching about Wo 2 oEni Fs 7 o. It 
may be presumed that, at the tempera- 
ture at which the two four-phase volumes 
Cpx + Opx + Trd + L and Opx + 01 + 
Trd + L have a common liquid, the 
reaction Opx + L -> Cpx + 01 + Trd 
will take place. It may be predicted that 
at a somewhat lower temperature where 
Cpx + Wo + Trd + L and Wo + 01 + 
Trd + L have a common liquid the 

Si0 2 

*---... Di.Hd 






Weight per cent 

Fig. 33. Schematic representation of relations at 1350°C along the line a-a' of figure 32 through 
the system CaO-MgO-FeO-Si0 2 . Fa, fayalite; Fo, forsterite; Kr, kirschsteinite; Mo, monticellite. 
Note that compositions of liquid are for the most part enriched in silica relative to the pyroxene plane. 



reaction of Trd + L will take place and 
yield Cpx + Wo + 01. It is likely that 
the last liquid will produce solid solutions 
of these three phases. 

The isothermal sections are, of course, 
based mainly on natural materials whose 
compositions cannot be fully represented 
by the pyroxene quadrilateral. For this 
reason the results illustrated are not in 
complete agreement with those that can 
be predicted from the bounding joins 
studied using pure synthetic compounds. 
The general phase relations and sequence 

of events, however, are in accord with 
those occurring in the bounding joins. 

Applications of these diagrams to 
igneous rock problems are manyfold, but 
discussion of the principal pyroxene 
trends is delayed until the final products 
of crystallization are ascertained. It will 
be of interest to correlate the behavior of 
some of the pyroxenes with the thermal 
studies of Tilley, Yoder, and Schairer on 
the rocks from which they have been 

Si0 2 

1350 °C 



Fig. 34. Schematic representation of relations at 1350°C for the MgO-FeO-Si0 2 system. Note 
that compositions of liquid are enriched in olivine components relative to the pyroxene plane (cf. 
Bowen and Schairer, 1935, p. 177, fig. 13). 



Pwo + Wo 


Trd + Opx+OI- 

Weight per cent 

-Trd + Opx+OI + L 

Fig. 35. Preliminary 1250°C diagram for compositions in the pyroxene plane of the CaO-MgO- 
FeO-Si0 2 system. Relations along the section a-a'-a" are portrayed in figure 36. Abbreviations as 
in figure 32. 




250 °C 



Fig. 36. Schematic representation of relations at 1250°C along the lines a-a' and a' -a" in figure 35 
through the CaO-MgO-FeO-Si0 2 system. Abbreviations as in figures 32 and 33. Note the position 
of the all-liquid region relative to the pyroxene plane. 





Weight per cent 

Fig. 37. Preliminary 1150°C diagram for compositions in the pyroxene plane of the CaO-MgO- 
FeO-Si0 2 system. Relations along the lines a-a' and a' -a" are portrayed in figure 38. Abbreviations 
as in figure 32. 



Opx +Trd 



Fig. 38. Schematic representation of relations at 1150°C along the lines a-a' and a' -a" of figure 37 
for the CaO-MgO-FeO-Si0 2 system. Abbreviations as in figures 32 and 33. Note that the all-liquid 
region is now entirely below the pyroxene composition plane. 






Weight per cent 

Fig. 39. Preliminary 1050°C diagram for compositions in the pyroxene plane of the CaO-MgO- 
FeO-Si0 2 system. Abbreviations as in figure 32. 



The Join Diopside-Silica 

J. F. Schairer and I. Kushiro 

The join diopside-silica, which was 
studied by Bowen (1914), has been re- 
examined to determine whether it is 
binary. If binary, it is a thermal barrier 
separating liquids that trend toward the 
MgO-Si0 2 join from those that trend 
toward the CaO-Si0 2 join in the system 
CaO-MgO-Si0 2 as suggested by Schairer 
and Yoder (Year Book 61, pp. 75-82). If 
it is not binary, it would not be a thermal 
barrier, and the liquids would change 

their compositions across it. The results 
obtained by the present experiments 
indicate that this join is not binary. 

The revised equilibrium diagram of the 
join diopside-silica is shown in figure 40. 
The liquidus minimum which is a piercing 
point exists at a composition between 
silica 15 and 16 weight per cent, which is 
almost identical with that obtained by 
Bowen (1914). The temperature of the 
piercing point is 1371° ± 1°C, about 10° 
higher than that of Bowen. 

The most significant fact is that there 
is a region of three-phase assemblage 




o 1380 




Tr + L 

CaMgSi 2 6 

10 15 20 

Weight per cent Si0 2 

25 -*-Si0 5 

Fig. 40. Revised equilibrium diagram of the diopside-rich portion of the system diopside 
(CaMgSi 2 6 ) -silica. Di B8 , diopside solid solution; L, liquid; Tr, tridymite. 



shown by the Di ss + Tr + L region in 
figure 40. This fact indicates that the join 
is not binary. The temperature interval 
of this region is about 25°, except for 
compositions rich in diopside. In the 
diopside-rich part of the join, the temper- 
ature at which diopside is joined by 
tridymite progressively decreases as the 

diopside content of the mixture increases. 
The X-ray powder patterns of diopside 
solid solutions coexisting with liquid or 
liquid and tridymite are different from 
those of pure diopside, although their 
differences are not large. A20[20(311) — 
20(310)] values of the diopside solid solu- 
tions are shown in table 7. All the A20 

CaSi0 3 


CaMgSi 2 6 

MgSi0 3 

Fig. 41. Three-phase triangles at 1370°, 1360°, and 1350°C. Boundary curves are from Bowen 
(1914), Schairer and Yoder {Year Book 61, p. 76), and Kushiro and Schairer (Year Book 62, p. 100). 
Di 8S , diopside solid solution; L, liquid; Fo S8 , forsterite solid solution; Pr ss , protoenstatite solid solution; 
Tr, tridymite; Wo, wollastonite. 

TABLE 7. A20 [20(311) - 20(310)] Values and Estimated Compositions 
of Diopside Solid Solutions in the System Diopside-Silica 

Composition of Mixture, 



of Run, °C 

A20[20(311) - 20(310)], 

Content of MgSi0 3 , 

wt. % 




0.615 ± 0.005 




0.586 ± 0.009 
0.606 ± 0.011 

4.5 ± 1.5 
2.0 ± 1.5 




0.600 ± 0.005 
0.572 ± 0.005 

2.5 ± 1.0 
6.5 ± 1.0 




0.601 ± 0.006 
0.564 ± 0.007 

2.5 ± 1.0 
7.5 ± 1.5 




0.598 db 0.010 
0.583 ± 0.010 

3.0 ± 1.5 
5.0 ± 1.5 



values are smaller than that of pure 
diopside. A possible solid solution that 
could explain the small A20 value is a 
diopside solid solution containing ensta- 
tite. The compositions of diopside solid 
solutions are estimated by the A20 value, 
which changes systematically from diop- 
side to diopside 65, enstatite 35, per cent 
(Kushiro and Schairer, Year Book 62, p. 
99). The maximum content of enstatite 
so far measured is 7.5 ± 1.5 weight per 
cent. It is to be noticed that the diopside 
solid solutions crystallized at higher 
temperatures contain more enstatite than 
those crystallized at lower temperatures, 
except one crystallized from the mixture 
diopside 96, silica 4, per cent. 

The probable three-phase triangles at 
1370°, 1360°, and 1350°C are shown in 
figure 41, on the basis of the data on the 
join diopside-silica and the compositions 
of the diopside solid solutions in table 7. 
As is shown by the progressive change 
of the three-phase triangles, the liquids 
in the system diopside-enstatite-silica 
change their compositions toward the 
invariant point diopside, tridymite, wol- 
lastonite, and liquid, across the join 
diopside-silica. This is consistent with the 
new results on the diopside ss -tridymite 

The precise measurements of the tem- 
peratures of the diopside ss -tridymite 
boundary indicate that no minimum or 
maximum exists along this boundary, and 
consequently the boundary drops contin- 
uously from the reaction point B to the 
join diopside-silica and further to the 
invariant point diopside, tridymite, wol- 
lastonite, and liquid. Recently, D. H. 
Speidel (oral communication, January 
1964) studied the diopside-tridymite 
boundary and found that its temperature 
drops continuously from B to the join 
diopside-silica, although the decrease is 
very small. 

The Join Diopside-Akermanite 

/. Kushiro and J. F. Schairer 

The join diopside-akermanite has been 
studied at 1 atm to determine the nature 

of solid solution occurring in diopside. In 
the studies of the join diopside-forsterite, 
we found that diopside contains about 5 
weight per cent forsterite in solid solution 
at least at temperatures higher than 
1300°C and that the melting point of the 
diopside solid solution containing 5 per 
cent forsterite is 2° higher than that of 
pure diopside (Kushiro and Schairer, 
Year Book 62, pp. 95-96). These results 
suggest the possibility that diopside may 
contain a small amount of akermanite in 
solid solution. The results obtained by the 
present experiments strongly suggest that 
diopside contains several weight per cent 
akermanite in solid solution at tempera- 
tures near the beginning of melting. 

The eutectic point was determined to 
be at 41 weight per cent akermanite, 59 
per cent diopside (fig. 42) , which is almost 
identical with that obtained by Ferguson 
and Merwin (1919). The temperature of 
this point is 1361° ± 1°C, about 6° lower 
than that of Ferguson and Merwin. 

The existence of diopside solid solution 
containing a small amount of akermanite, 
shown by the area Di ss in figure 42, was 
found by the same procedures as those 
followed in the study of the join diopside- 
forsterite. The X-ray powder pattern of 
the mechanical mixture 95 weight per 
cent synthetic diopside, 5 weight per cent 
synthetic akermanite shows distinct re- 
flections of akermanite. After heating the 
mechanical mixture at subsolidus tem- 
peratures, 1355° and 1360°C, for 3 days, 
these reflections disappeared. A glass of 
the same composition was also crystal- 
lized only as diopsidic pyroxene at 1355° 
and 1360°C for 18 hours. This evidence 
strongly indicates that a small amount of 
akermanite dissolved in diopside in solid 
solution. Another mechanical mixture 
consisting of 90 per cent diopside and 10 
per cent akermanite was heated at 1355°C 
for 2 days. After heating, the intensities 
of the X-ray reflections of akermanite 
relative to those of diopside were much 
reduced and were nearly the same as those 
of the unheated mechanical mixture con- 
taining 5 per cent akermanite. This 







a. 1380 






i i 



I l 

J 1 





Ak + L 



Di ss +L 





/ o 


- / 






1 I 



I ! 

1 1 



CaMgSi 2 6 

20 30 40 50 60 70 

Weight per cent Ca 2 MgSi 2 7 


90 Ca 2 MgSi 2 7 

Fig. 42. Revised equilibrium diagram of the system diopside (CaMgSi 2 6 )-akermanite 
(Ca 2 MgSi 2 07). Ak, akermanite; Di 8S , diopside solid solution; L, liquid. 

evidence indicates that about 5 per cent 
akermanite dissolved in diopside in solid 
solution at 1355°C. The limit of the solid 
solution would be reduced with decreasing 
temperature. At 1350° and 1300°C, both 
the mechanical mixture and glass of the 
composition 95 per cent diopside and 5 
per cent akermanite crystallized into 
diopside and akermanite. At magmatic 
temperatures only a very small amount 
of akermanite would be expected in solid 
solution in diopside. 

Deduction of Liquid Crystallization 

Paths in a Five-Component Oxide 

System Containing Iron 

D. C. Presnall 

The representation of compositional 
relations in a system of more than four 

components is a complicated geometrical 
problem. In general, a single diagram is 
insufficient for this purpose. However, it 
can be shown that a series of diagrams 
may be used to illustrate phase relations 
quantitatively and in a legible form for a 
five-component system containing iron in 
varying degrees of oxidation. Such repre- 
sentations are particularly useful, for they 
would reduce the uncertainty in applying 
experimental data on simplified systems 
to problems of magmatic differentiation. 
In the study of iron-bearing ternary 
and quaternary oxide systems at a 
constant total pressure of 1 atm, it is 
common to remove one degree of freedom 
by fixing the oxygen pressure of the gas 
in equilibrium with the condensed phases 
(Muan and Osborn, 1956; Muan and 
Somiya, 1960) or by fixing the mixing 



ratio of two gases such as C0 2 and H 2 
(Muan and Osborn, 1956). Equilibrium 
data from such studies are frequently 
represented by showing the phase-equi- 
librium relations on a chosen join in 
which the iron is in an arbitrary fixed 
state of oxidation. For example, Muan 
and Osborn (1956) found a variable 
Fe 2 3 -FeO ratio in liquids of the system 
MgO-iron oxide-Si0 2 in equilibrium with 
air. For purposes of representation, they 
converted all iron to Fe 3 4 , recalculated 
the oxide percentages to 100 per cent by 
weight, and plotted the resulting compo- 
sitions on the join MgO-Fe 3 4 -Si0 2 . This 
is equivalent to projecting all liquid 
compositions onto the join MgO-Fe 3 4 - 
Si0 2 along lines defined by the compo- 
sitions in question and the oxygen apex 
of the tetrahedron Mg-Fe-Si-O. These 
lines are the "oxygen reaction lines" of 
Muan (1958, p. 201) and the "total 
composition lines" of Osborn (1959, p. 
613). Equilibrium relations at constant 
oxygen pressure in a quaternary system 
such as this have the appearance of a 
ternary equilibrium diagram when so 

Muan and Somiya (1960), in their 
study of the system iron oxide-Cr 2 3 - 
Si0 2 in equilibrium with air, also chose to 
calculate all iron oxide as Fe 3 4 . They 
then deduced crystallization paths at 
constant oxygen pressure directly from 
the triangular diagram Fe 3 4 -Cr 2 3 -Si0 2 
just as if it were the equilibrium diagram 
of a ternary system and without referring 
to the tetrahedron FeO-Fe 2 3 -Cr 2 3 -Si0 2 . 
Muan and Somiya did not determine the 
deviations of the compositions of their 
mixtures from the plane Fe 3 4 -Cr 2 3 - 
Si0 2 . Nevertheless, their treatment of 
crystallization paths at constant oxygen 
pressure is rigorously correct, since the 
total composition of a mixture, as it 
shifts along its total composition line 
during cooling, remains at the same point 
in space when all iron oxide in the mixture 
is recalculated to Fe 3 4 . That is, the 
starting composition, in terms of Si0 2 , 

Cr 2 3 , and total iron oxide as Fe 3 4 , does 
not move on the join Fe 3 4 -Cr 2 3 -Si0 2 as 
the oxygen content of the mixture varies 
during cooling. Thus, all the usual 
geometric devices for deducing crystal- 
lization paths in a ternary system can be 

For a four-component oxide system 
containing iron all the liquidus equilib- 
rium relationships at a constant oxygen 
pressure can be shown on a projected 
diagram such as that used by Muan and 
Somiya (Presnall, 1963). In order to 
reclaim the information lost during pro- 
jection, contours of the Fe 2 3 -FeO ratios 
of liquids (Muan and Osborn, 1956) and 
of crystalline phases in equilibrium with 
these liquids are drawn on the liquidus 
surface of the projected diagram. When 
all this information is shown, true propor- 
tions of phases in equilibrium can be 
determined by first measuring the appa- 
rent relative proportions of phases from 
the projected diagram and then recalcu- 
lating the apparent proportion of each 
phase to its true proportion, using the 
contoured information on the Fe 2 3 -FeO 
ratios of the various phases. The calcu- 
lation of the oxygen content of the total 
mixture follows immediately. 

When the above treatment is extended 
to a five-component oxide system contain- 
ing iron, the equilibrium relationships at 
a constant oxygen pressure are repre- 
sented in a tetrahedron, total iron oxide 
being calculated as, say, FeO, and plotted 
at one apex (Presnall, 1963). Again, no 
compositional information at the constant 
oxygen pressure need be sacrificed by 
projecting into the tetrahedron. The data 
on Fe 2 3 -FeO ratios of the various phases 
in equilibrium at liquidus temperatures 
would be shown by contour surfaces 
passing through the various primary 
phase volumes. The projected constant 
oxygen pressure diagram would look like 
a quaternary system and could be treated 
as such when deducing crystallization 
paths at constant oxygen pressure. As 
before, true relative proportions of phases 



could be calculated using the contoured 
data on Fe 2 3 -FeO ratios of phases. 

Crystallization at constant oxygen 
pressure is often an unrealistic approxi- 
mation to natural conditions during the 
crystallization of a magma. When apply- 
ing the experimental data to the crystal- 
lization of magmas, it is therefore 
desirable to be able to deduce liquid 
crystallization paths when the total 
composition of the starting mixture is 
fixed, that is, when there is no exchange 
of oxygen between the system and its 
surroundings. Under these conditions the 
oxygen pressure changes continuously 
during cooling. 

For a five-component oxide system 
containing iron all the equilibrium rela- 
tions at a constant oxygen pressure can 
be represented in a tetrahedron, as 
described above. If the system were 
studied at a sufficiently large number of 
oxygen pressures, the entire five-compo- 
nent system could be mapped and shown 
as a series of constant oxygen pressure 
tetrahedra, each of which would appear 
just as if it were the equilibrium diagram 
of a quaternary system. Now, for a 
system studied at a total pressure of 1 
atm, if the total composition and temper- 
ature are given the phases in equilibrium 
and the compositions of these phases are 
completely defined. Also, the relative 
proportions of the phases in equilibrium 
are defined except at an invariant situ- 
ation. Therefore, the five-component 
system having been completely mapped, 
if a total composition and temperature 
are given it should be possible to find the 
equilibrium assemblage of phases and the 
proportions of these phases. To deduce a 
crystallization path, successively lower 
temperatures are assigned for this compo- 
sition, and the equilibrium assemblage 
is found for each temperature. This pro- 
cedure gives a stepwise crystallization 
history with steps spaced as closely as the 
data permit. 

In more detail, the procedure is as 
follows: Consider the system A-B-C- 

Fe 2 3 -FeO. Determine a series of liquidus 
equilibrium diagrams for this system, 
each at a different oxygen pressure. Plot 
the diagrams as tetrahedra with A, B, C, 
and total iron oxide as FeO at the apices. 
Draw a series of isothermal sections for 
each constant oxygen pressure tetra- 
hedron. For each isothermal tetrahedron 
at a constant oxygen pressure, the 
Fe 2 3 -FeO ratios of compositions in the 
tetrahedron will vary. These ratios can be 
experimentally determined and shown as 
contour surfaces passing through the 
various one-, two-, three-, and four-phase 
fields in the isothermal tetrahedron. 
When the task is completed, there will be 
a series of isothermal tetrahedra for each 
oxygen pressure. Now take an arbitrary 
starting composition in the five-compo- 
nent system. This composition, except for 
the Fe 2 3 -FeO ratio, can be plotted in 
each constant oxygen pressure tetra- 
hedron. Then choose a temperature for 
which a series of isothermal sections has 
been drawn as a function of oxygen 
pressure. In general, there will be one 
oxygen pressure at which the Fe 2 3 -FeO 
ratio of the mixture is the same as that 
initially chosen. Thus, if isothermal sec- 
tions at the chosen temperature have been 
studied at sufficiently close intervals of 
oxygen pressure, there will be one section 
that shows a contour surface of Fe 2 3 - 
FeO ratio exactly passing through the 
point corresponding to the total compo- 
sition of the mixture plotted in terms of 
A, B, C, and total iron oxide as FeO. This 
section will then yield the compositions of 
the phases in equilibrium and (except at 
an invariant point) their relative propor- 
tions when all iron is calculated as FeO. 
By means of the contoured information 
on the Fe 2 3 -FeO ratios of the various 
phases, the apparent proportions of the 
various phases can be recalculated to 
their true proportions. Repetition of this 
procedure at a large number of tempera- 
tures gives the crystallization history of 
the mixture. 



Upper Stability Limits of 

Magnesian Chlorites 

J. J. Fawcett 

Studies of the upper stability limits of 
magnesian chlorites reported in Year Book 


62, page 140, have been continued, and 
the curve has been revised with respect 
to the preliminary data presented in that 
report. The breakdown curve for the 
magnesian chlorites passes through the 
points 768° =fc 7°C at 3.5 kb P H .o, 787° 




*— r- 

y^ p- , 

i i i 




>- + 









x |x XX 


Chi — 

x/x x 

/ x " 




7 Co X 

X X 

x Po + Co 



Vr-Beginning of melting 


X V X 

^ \ 


X \x 









i i 



. 1 I \a 






I 100 1200 1300 1400 

Temperature , °C 

Fig. 43. The upper stability limits of the magnesian chlorites to 10 kb Ph 2 o. Point A is taken 
from the data of Rankin and Merwin (1918), and below 2 kb Ph 2 o the curve is taken from Yoder 
(1952) for the upper stability limit of clinochlore. 



=b 7°C at 5 kb P Ha o, and 830° ± 5°C at 
10 kb Ph 2 o- As was noted last year, the 
breakdown products of magnesian chlo- 
rite vary with pressure. The assemblage 
forsterite + cordierite + spinel was ob- 
tained from magnesian chlorites by Yoder 
(1952) and Roy and Roy (1955). At high 
water pressures, however, the forsterite 
-f- cordierite assemblage is no longer 
stable, and the breakdown assemblage of 
magnesian chlorites is forsterite + ensta- 
tite + spinel. The revised values for the 
invariant point at which forsterite, ensta- 
tite, spinel, cordierite, chlorite, and vapor 
coexist are 765° db 10°C and 3.25 kb ?h 2 o 
± 0.25 kb. 

The relationship between the high- and 
low-pressure breakdown assemblages is 
very important, as the minerals forsterite 
and cordierite are rarely found together 
in nature whereas the assemblage ensta- 
tite + spinel is common in both igneous 
and metamorphic rocks. It has not been 
possible to place narrow brackets on the 
curve for the reaction enstatite + spinel 
+± forsterite + cordierite, owing to the 
slow rates of reaction, even at high 
temperatures. For example, the assem- 
blage forsterite + enstatite + spinel has 
been completely converted to forsterite + 
cordierite + spinel at 1100°C and 2 kb 
Ph 2 o in 96 hours, and the reaction has 
been reversed at 1100°C and 3 kb P Hs o in 
39 hours. At the same temperature, how- 
ever, and the intermediate pressures of 
2.5 and 2.75 kb Ph 2 o, neither assemblage 
showed any sign of reaction after 95 
hours. The available data suggest that the 
curve may have a slight negative slope 
(fig. 43). 

Points bracketing the chlorite upper 
stability curve have been replotted in 
terms of fugacity and 1/T °K. The slope 
of the curve thus obtained is — AH/2.303R 
(from the integrated Clausius-Clapey- 
ron relationship), and thus AH for the 
reaction can be evaluated. For the reac- 
tion magnesian chlorite — > enstatite + 
forsterite + spinel, AH is 73 kcal/mole in 
contrast to 24 kcal/mole for the low- 
pressure reaction magnesian chlorite — > 

forsterite + cordierite + spinel (from the 
PT data of Yoder, 1952). 

The Muscovite-Chlorite-Quartz 


J. J. Fawcett 

The mineral assemblage muscovite- 
chlorite-quartz is most commonly found 
in rocks of the greenschist facies of 
regional metamorphism, and the phase 
relations between the minerals are of 
considerable importance in the concept 
of progressive metamorphism. It is upon 
a foundation of the physical and chemical 
properties of these minerals that the 
sequence from the chlorite zone to biotite, 
garnet, and amphibolite zones is based in 
many interpretations of chemical reac- 
tions of regional metamorphism. A precise 
knowledge of the phase relations among 
the various minerals is thus of prime 
importance if we are to determine the 
conditions and processes of metamor- 
phism. Relationships between magnesian 
chlorites and quartz have been clarified 
during the last two years (Year Book 61, 
pp. 88-91; Year Book 62, pp. 139-143) by 
studies in the system MgO-Al 2 03-Si0 2 - 
H 2 using runs of much greater duration 
than those of earlier workers. As several 
studies of the phase relations of muscovite 
and its breakdown products have now 
been made (see Velde, this report), it is 
pertinent to examine the assemblage 
muscovite + chlorite + quartz, although 
an alternative and equally important 
approach would be an investigation of the 
chlorites of intermediate Mg/Fe ratio 
such as those most commonly found in 

Among others, Goldschmidt (1921), 
Tilley (1924), Harker (1939), Turner and 
Verhoogen (1951), Yoder (1959), and 
Ernst (1963) have commented on the 
presence of muscovite and chlorite in 
low-grade metamorphic rocks and their 
relationship to the production of biotite 
at the biotite isograd. Several reactions 
have been suggested for the removal of 
chlorite from low-grade metamorphic 



rocks, but in spite of the fact that it may 
be possible to balance the equations there 
is sometimes little else to support the 
application of the reaction to natural 
processes. The most frequently cited re- 
actions center around the varying compo- 
sitions of the phases involved (table 8). 
Thus a siliceous muscovite (phengite) 
may react with chlorite to produce mus- 
covite, biotite, and quartz (reaction 1); 
an alumina-free chlorite may react with 
muscovite, producing biotite, aluminous 
chlorite (corundophilite), and quartz 
(reaction 2); or muscovite and chlorite 
may react to give biotite, andalusite, and 
quartz (reaction 3). Alternatively, pure 
muscovite itself may react with chlorite 
to produce an aluminous biotite, alumi- 
nous chlorite, and quartz (reaction 4). 

The most common substitutions in the 
natural analogues of these idealized reac- 
tions are Fe 2+ for Mg and Fe 3+ for Al. 
The nature of the reactions and the 
resulting assemblages are clearly depend- 
ent on the proportions of muscovite and 
chlorite in rocks comprising the green- 
schist facies. Subsequent stages of meta- 
morphism involve the production of 
garnet, but the physical conditions under 

which garnet forms will differ according 
to the nature of the starting materials. 
Thus, if chlorite has been completely 
consumed in order to produce biotite, the 
appearance of garnet will be inhibited 
until a temperature is reached at which it 
is produced by a reaction involving 
biotite rather than chlorite. In the pres- 
ence of excess chlorite, however, garnet 
may begin to form at an earlier stage 
owing to reactions involving chlorite. 

A series of experiments has been de- 
signed to examine the phase relations of 
the muscovite + chlorite + quartz 
assemblage and its bearing on the produc- 
tion of biotite and garnet in rocks of the 
biotite and garnet zones of progressive 
regional metamorphism. The initial ex- 
periments have been carried out with 
iron-free natural and synthetic powders 
for starting materials, and thus the phase 
relations may be depicted in the system 
K 2 0-MgO-Al 2 3 -Si0 2 -H 2 0. Figure 44 
shows the relationships among some of 
the phases in that system. As none of the 
phases of interest is highly potassic it is 
not necessary to illustrate that part of the 
system. One of the apices is therefore 
designated K 2 + A1 2 3 in the ratio 1:1, 


Chemical Reactions Possible in Rocks in the Greenschist Facies 

(1) 8(K 2 0-Mg0.2Al 2 3 -7Si0 2 -2H 2 0) + 2(5MgO -Al 2 3 -3Si0 2 . 4H 2 0) ->5(K 2 0- 3 Al 2 3 -6Si0 2 -2H 2 0) 

Phengite Clinochlore Muscovite 

+ 3(K 2 0-6MgO-Al 2 3 -6Si0 2 -2H 2 0) + 14Si0 2 + 8H 2 

Biotite Quartz 

(2) 6Mg0.4Si0 2 -4H 2 + K 2 • 3A1 2 3 • 6Si0 2 • 2H 2 -> K 2 • 6MgO • A1 2 3 • 6Si0 2 • 2H 2 

Talc-Cklorite Muscovite Phlogopite 

+ 4MgO-2Al 2 3 -2Si0 2 .4H 2 + 2Si0 2 

Corundophilite Quartz 

(3) 10(K 2 O-3Al 2 O 3 -6SiO 2 -2H 2 O) + 6(5MgO-Al 2 3 -3Si0 2 -4H 2 0) 

Muscovite Clinochlore 

-» 10(K 2 O-6MgO-Al 2 O 3 -6SiO 2 -2H 2 O) + 16Al 2 Si0 6 + 2Si0 2 + 24H 2 

Phlogopite Andalusite Quartz 

(4) K 2 0-3Al 2 3 -6Si0 2 -2H 2 + 5MgO • A1,0 3 • 3Si0 2 ■ 4H 2 -* 4MgO-2Al 2 3 -2Si0 2 -4H 2 

Muscovite Clinochlore Corundophilite 

+ K 2 0-5MgO-2Al 2 3 -5Si0 2 -2H 2 + 2Si0 2 

Eastonite Quartz 



and both phlogopite and celadonite lie on 
the face K 2 0:Al 2 03-MgO-Si02. As mus- 
covite lies on the back face of the tetra- 
hedron, the plane muscovite-phlogopite- 
quartz extends from the left front face to 
the back face of the tetrahedron. The 

muscovite- celadonite solid solution series 
lies to the left of that plane, and so any 
tie lines from muscovite solid solutions to 
chlorite solid solutions penetrate that 

Another relationship illustrated in this 

Si0 2 




Fig. 44. Relationship of some of the phases inf the system K 2 0-MgO-Al 2 03-Si02-H 2 0. Note, 
however, that the tetrahedron shows only part of this system, as one of the apices represents K 2 + 
AI2O3 in the ratio 1:1 and the diagram assumes excess H 2 0. The phases phlogopite (Ph) and cela- 
donite (CELAD) plot on the left front face of the tetrahedron, and, as muscovite (Ms) plots on the 
back face, the plane Ph-Si0 2 -Ms extends from the front to the back of the tetrahedron. Fine dotted 
lines illustrate the solid solution in the various phases, and the phases clinochlore (CI) and amesite 
(Am) plot on the right front face of the figure. The line joining amesite and celadonite illustrates 
the spatial relationship of that assemblage to the Ms-Ph-Si0 2 assemblage. 



figure is that of the potassium feldspar -f 
chlorite assemblage. The tie line joining 
chlorite solid solution compositions to 
alkali feldspar must clearly penetrate the 
plane muscovite-phlogopite-quartz. Al- 
though the assemblage chlorite + alkali 
feldspar is rarely reported in rocks from 
the greenschist facies, it is more common 
in the zeolite facies and in sedimentary 
rocks (Dickinson, 1962; Harrison and 
Campbell, 1963). There has, however, 
been at least one report of the coexistence 
of these minerals in the greenschist facies 
(Zen, 1960), and Chayes (1955) discussed 
the production of potash feldspar as a 
product of the biotite- chlorite transfor- 
mation. It may prove possible to regard 
the appearance of the alkali feldspar + 
chlorite assemblage as a distinctive 
isograd in the passage from the green- 
schist to the zeolite facies. 

The diagrammatic presentation of data 
obtained from the compositions studied 
is somewhat complex; the data cannot be 
represented on a simple diagram. The 
projection used to illustrate phase rela- 
tions in the system K 2 0-MgO-Al 2 3 -Si0 2 - 
H 2 is shown in figure 45, in which the 
apices of the triangle are K+, Mg 2+ , and 
Al 3+ , and the projection assumes excess 
H 2 and Si0 2 . Thus both quartz and a 
water vapor phase exist in equilibrium 
with each phase assemblage indicated. 
The projection can be modified to the 
more general diagram necessary to ac- 
commodate natural assemblages by as- 
signing the coordinates K + Na, — R 3+ , 
— R 2+ . Because of its inherent simplicity 
this plot illustrates results of work on 
synthetic materials better than other 
projections do. Various starting materials 
have been tried in runs to determine 
which is most suitable to promote rapid 
reactions and also avoid, whenever possi- 
ble, the production of metastable phases. 
Thus glasses, oxide mixtures, gels, and 
natural and synthetic minerals have all 
been used as starting materials. In 
general, natural minerals are unsuitable, 
as they persist metastably for many 

weeks at low pressures and temperatures 
(i.e., 2 kb P Hj o below 625°C). Since 
glasses and gels are most prone to the 
production of metastable phases, the 
synthetic minerals themselves are most 
suitable, either individually or mixed in 
known proportions, as starting materials. 
The best method for confirming the 
pressure and temperature of a reaction is 
to demonstrate the incompatibility on 
various mineral assemblages on either 
side of a boundary curve. 

Velde's work on the join muscovite- 
celadonite (this report) demonstrates the 
critical nature of muscovite solid solutions 
in controlling the phase assemblages. The 
wide range of alumina content possible in 
chlorite compositions further complicates 
the phase relations of the various assem- 
blages. A mixture of natural muscovite 
and clinochlore was held at 350°, 450°, 
and 550°C and 2 kb Ph 2 o for 9 weeks 
without any sign of reaction. Although 
natural minerals are slow to react, the 
absence of any visible signs of reaction 
suggests that they constitute a stable 
assemblage under the conditions investi- 
gated. Additional runs at 2 kb P Hs o on 
mixtures containing various proportions 
of sanidine and clinochlore have thus far 
failed to locate the equilibrium conditions 
for this assemblage. A mica, presumably 
a muscovite solid solution, together with 
K feldspar and chlorite was produced 
from the initial feldspar-chlorite starting 
material. Although the three-phase assem- 
blage K feldspar + mica + chlorite (less 
aluminous than the initial clinochlore) 
could well be stable under these condi- 
tions, the results so far are inconclusive. 
Additional runs of greater duration are 
necessary to allow more complete reaction 
for a variety of starting materials. Pre- 
liminary phase relations at 450°C and 2 
kb ?h s o are shown in figure 45. 

A supplementary approach to increas- 
ing the duration of low-pressure runs is 
to determine the phase relations at high 
pressures (5 and 10 kb Ph 2 o) and project 
them to low pressures, reducing, it is 



hoped, the number of long runs required 
at low pressures. Preliminary experiments 
suggest that reactions requiring several 
weeks at 2 kb Ph 2 o niay be almost 
completed after 48 hours at 10 kb Pn 2 o- 
In the present work, however, the effect 
of pressure on the muscovite solid solution 
series (Velde, this report) will be a com- 
plicating factor. Mixtures of sanidine and 
clinochlore have been completely con- 
verted to chlorite + muscovite solid 

solution at 325° and 400°C, 10 kb P Ha o, 
in runs of 2 to 3 days. This approach will 
now be pursued in detail to determine 
phase relations of muscovite, chlorite, 
phlogopite, and quartz in the system 
K 2 0-MgO-Al 2 3 -Si0 2 -H 2 0. 

Upper Stability of Muscovite 
B. Velde 

The upper stability of muscovite was 
determined at five pressures (Ptotai ~ 

2kbP H; ,o450 C 

Mol per cent 

Fig. 45. Projection of phase relations into the diagram K + -Al 3+ -Mg 2+ to illustrate the relation- 
ships between muscovite solid solution, chlorite solid solution, phlogopite, and potash feldspar at 
450°C and 2 kb Ph 2 o. Phlogopite solid solutions also occur but are omitted here for clarity and for 
lack of any precise data on the phlogopite-eastonite (Ea) join in this system. The projection assumes 
excess water vapor and excess Si0 2 . Crosses indicate compositions of the starting materials, and the 
extent of Ms solid solution is taken from Velde (this report). 



i°H 2 o). The reaction muscovite £± sani- 
dine + corundum was observed in both 
directions as indicated in the equation. 
Products were determined on the basis of 
X-ray and optical observations, the 
optical method being found necessary for 
observing the breakdown of muscovite. 
The presence of the desired product in 
amounts of about 15 per cent was assumed 
as an indication of stability. Three 
materials were used as reactants: syn- 
thetic muscovite (produced by the reac- 
tion kaolinite + KOH — > muscovite), a 
natural muscovite (properties given by 
Velde, 1964), and synthetic sanidine + 
corundum (produced by dehydrating 
natural muscovite at 900°C and 1 atm 
for 24 hours). The brackets determined 
for the interval over which the reaction 
took place and the estimated precision of 
the temperature measurements are given 
in figure 46 (Ph 2 o ~ Ptotai). Curves given 
by Yoder and Eugster (1955) and Crowley 

and Roy (1964) are also shown. Calcu- 
lation of the relation 

This work 

Sanidine + 

d log /h 2 o 

AH ( 


Fig. 46. Upper stability of muscovite, 
Ph 2 o ^ Ptotai- C & R, Crowley and Roy (1964); 
Y & E, Yoder and Eugster (1955). 

by graphical methods gives AH° = ^73 
kcal/mole. / = fugacity of water, T in 
degrees Kelvin. Results of the present 
study and those of Crowley and Roy 
(1964) and Yoder and Eugster (1955) are 
discussed elsewhere (Velde, 1964); it 
appears, however, that the experimental 
method may have caused the variation 
in results. 

Low-Grade Metamorphism of 
Micas in Pelitic Rocks 

B. Velde 

Dioctahedral Micas and Related 
Mineral Groups 

One of the major problems in the study 
of the pelitic components of sedimentary 
rocks is the genesis of the dioctahedral 
mica- like minerals present. These micas, 
having a basic (001) spacing of about 10 
A, are generally called illite and are 
abundant in sedimentary rocks. Yet little 
is known about the possible range in 
composition that can occur in the mineral 
group, nor are there adequate data on the 
requisite conditions for their occurrence. 
It has been fairly well established that 
the major proportion of illite found in 
sedimentary rocks is of diagenetic origin. 
The importance of the illite mineral group 
to geologists is evident from an assess- 
ment of the mineral constituents of rocks 
at the earth's surface. Illite is the most 
abundant pelitic mineral in sedimentary 
rocks, accounting roughly for two-thirds 
of these minerals. Shales and shaly rocks 
comprise three-fourths of the sedimentary 
rocks, which represent about three- 
fourths of the rocks at the earth's surface 
(Pettijohn, 1957). Illite, then, is one of 
the most abundant minerals in sedimen- 
tary rocks and provides a basic material 
for most low-grade metamorphic mineral 
reactions. Therefore an understanding of 
the behavior of this mineral group under 



various PT conditions will greatly aid in 
studying the petrology of sedimentary 
and low-grade metamorphic rocks. 

In most dioctahedral micas potassium 
is the major interlay er ion present in the 
structure. These micas have four cations 
in sixfold and two in eightfold coordina- 
tion with oxygen and hydroxyl anions. 
The basic (001) spacing is near 10 A. 
This description includes illites; it also 
includes several other natural micas. 
Foster (1956) has classified dioctahedral 
mica compositions in the trisilicic to 
tetrasilicic series: 

M+(R+ 3 ) 2 (Si 3 R+ 3 )O 10 (OH) 2 - 

M+(R+ 2 R+ 3 )Si 4 Oio(OH) 2 , 

where M+ = K, Na, Ca, or H 3 0+; R+ 2 = 
divalent metal ion; R+ 3 = trivalent metal 
ion. Considering natural mica compo- 
sitions this classification seems quite 
reasonable, and in fact a series 

muscovite (KAl 2 Si 3 A10io[OH] 4 )- 
celadonite (KR+ 2 R+ 3 Si 4 Oi [OH] 2 ) 

is possible. Natural muscovites rarely 
contain significant quantities of R +3 
ions that are not aluminum (Kanchira 
and Banno, 1960). In the present 
investigation 2-month hydrothermal 
experiments were run with mixtures 
of oxides (Si0 2| Fe 2 3 , and KOH) 
and kaolinite as starting materials 
in order to determine the amount of solid 
solution in the muscovite series : 

KAl 2 Si 3 AlO 10 (OH) 2 - 
KFe+ 3 2 Si 3 AlO 10 (OH) 2 . 

Only about 15 per cent of the iron mica 
could be accommodated by the muscovite 
structure in solid solution. "Ferri-musco- 
vite" was never produced. Therefore, the 
simplification of considering the diocta- 
hedral, potassic micas as a muscovite- 
celadonite system seems reasonable. Such 
a system has been investigated, consider- 
ing the four possible celadonite compo- 
sitions KAlMgSi 4 Oio(OH) 2 , KAlFeSi 4 Oio- 
(OH) 2 , KFe+ 3 MgSi 4 Oio(OH) 2 , KFe+ 2 Fe+ 3 - 
Si 4 Oio(OH)2, and aluminum muscovite. 

All runs were made in Stellite bombs 
using Ag 70 Pd 30 capsules, thus fixing the 
oxygen fugacity near the nickel-nickel 
oxide buffer pair (Eugster and Wones, 

Synthesis of micas within this system 
after runs of 1 month, and sometimes 2 
months, was limited to the conditions 
specified by the lined areas in figure 47. 
The most extensive solid solution extends 
from muscovite toward the MgAl cela- 
donite mica. The MgAl celadonite-musco- 
vite series was significantly pressure 
dependent; i.e., more solid solution 
appeared with increased water pressure 
(■Ptotai == Pn 2 o)- Aspects of this relation 
are dealt with later. The synthetic micas 
intermediate between muscovite and 
celadonite represent the natural phengitic 
micas (Ernst, 1963) and other micas 
associated with igneous occurrences (Fos- 
ter, 1956). Phengites are often associated 
with high-pressure minerals like jadeite 
and kyanite. It seems reasonable to com- 
pare the synthetic phengites to the 
natural micas because both appear asso- 
ciated with high pressures of formation. 
A natural phengite from Gran Paradiso, 
Italy (Michel, 1953), had the same PT 
stability as synthetic materials of similar 

The other area of the system in which 
micas were synthesized is essentially in 


4.5 kb 





Fig. 47. Compositions synthesized in the 
muscovite-celadonite system indicated by lined 
areas. Celadonite compositions are indicated by 
the various ion pairs present in the KR +2 - 
R +3 Si 4 Oio(OH) 2 composition. 



the plane of celadonite compositions, 
mainly between the MgFe +3 and Fe +2 Fe +3 
celadonites. Natural celadonite and glau- 
conites can be plotted in this general 
compositional area. These synthetic micas 
do not form increased solid solutions with 
muscovite under higher water pressures. 
Runs using natural glauconites and the 
synthetic glauconite-celadonite micas give 
a maximum thermal stability of about 
380°C. Above this temperature the syn- 
thetic and natural glauconites yield a 
biotite + quartz + iron oxide assemblage 
as found previously by Winkler (1964) in 
similar investigations of a natural glau- 

The definite gap between the compo- 
sitions of the synthetic phengite-type 
micas and the glauconite type (fig. 47) 
suggests that no continuous series exists 
between the glauconites, sedimentary in 
origin, and the micas of metamorphic 
rocks (phengites). Thus, within the mica 
compositions it appears impossible to 
transform a sedimentary mica into a 
muscovite through a metamorphic proc- 

There have been demonstrations 
(Winkler, 1957; the present study) that 
sedimentary illite will change its compo- 
sition under experimental hydrothermal 
conditions by the reaction illite — > mica + 
chlorite. This is quite compatible with 
previous field observations or a general 
consideration of low-grade metamorphic 
trends in pelitic rocks. The lower-grade 
greenschist facies consists of primarily 
mica-chlorite rocks. Illites have a general 
composition near that of glauconite (Jung, 
1954). Such a mineral could not produce 
a phengite or near muscovite mica by 
gradual change in composition if it is to 
remain within the mica system, since 
micas of intermediate composition are not 
stable under low-grade metamorphic 
conditions. This is seen in figure 47. Illite 
compositions are usually potassium (or 
Na,Ca) deficient and thus, from a chem- 
ical viewpoint, could not be considered 
micas. However, illites from Paleozoic 

rocks apparently have muscovite struc- 
tures as far as can be determined by 
X-ray diffraction studies (Velde and 
Hower, 1963). It must be concluded that 
illites represent a mica-like group of 
minerals but that they cannot be classi- 
fied as true dioctahedral potassic micas. 
Thus it appears that dioctahedral 
micas and related mineral groups can be 
separated into three groups: muscovites 
and phengites, which have strongly 
pressure- dependent stabilities and which 
approximate a muscovite composition; 
celadonite-glauconites, which have no 
pressure dependence in occurrence and 
are found in sedimentary or near surface 
environments; and the illites, which 
are intermediate, but not a part of the 
mica system, and which are probably 
antecedent to the phengites in prograde 
metamorphic sequences, because they 
have similar structures and compositions 
and appear to form micas in experimental 
studies using natural materials. 

Illite-Chlorite Relations at Low 
Temperatures and Pressures 

It has been noted (Grim, 1953; Weaver, 
1959) that the most common association 
of clay minerals in sedimentary rocks is 
that of illite and chlorite; accessory 
minerals are usually quartz and feldspar. 
This is similar to the muscovite-chlorite- 
quartz-feldspar assemblage of the green- 
schist facies in metamorphic rocks (Tur- 
ner, 1948). The association of mica and 
chlorite minerals is very important in 
sedimentary and low-grade metamorphic 
rocks because both these minerals can 
change composition significantly (substi- 
tutions of Mg for Fe+ 2 and Fe+ 3 , Al for 
Si). Feldspar and quartz are present in 
varying amounts but generally play 
subordinate roles in mineral transforma- 
tions of low-grade metamorphism. There- 
fore they are not diagnostic of the PT 
conditions that the rocks have experi- 
enced. Experiments by Winkler (1957) 
showed that a mica + chlorite assemblage 
can be produced from a natural illite. 



Using the knowledge gained from the 
study of the muscovite-celadonite system, 
it can be predicted that the mica in a 
mica-chlorite rock would be near musco- 
vite in composition under 1 and 2 kb 
pressures. At higher pressures, the mica 
could contain considerable amounts of 
MgAl celadonite in solid solution (up to 
80 per cent at 300°C and 10 kb). Investi- 
gations are being made in the muscovite- 
MgAl celadonite series using potassium- 
poor bulk compositions (^6 per cent K 2 
by weight, compared with muscovite, 
which has 11.8 per cent). These lower 
K 2 compositions can be considered part 
of a mica-chlorite system which assumes 
excess quartz and water. Results at 2 kb 
and 300° to 500°C thus far have shown 
that a one-phase product appears — a 
mixed layering of mica- and chlorite-like 
units. Such a mineral is analogous to the 
mixed-layer clay minerals often called 
illites. Through a study of the mineral 
associations produced by a variation in 
PT conditions it is believed that insight 
will be gained into the occurrence and 
behavior of illite under metamorphism. 

In studying the muscovite-celadonite 
systems at low temperatures and pres- 
sures, information has been gained about 
mixed-layer mica-montmorillonoids which 
are similar to structures found in illites. 
Results for one such muscovite-celadonite 
compositional series are shown in figure 
48. Variation in the number of mixed 
layers (montmorillonoid in mica) with 
temperature, pressure, and composition 
is important to note. Increase in tempera- 
ture with composition and pressure 
constant results in a product with fewer 
montmorillonoid layers in the mica 
structures. The same relation exists when 
pressure is increased with the remaining 
variables constant. Composition varia- 
tions from muscovite to celadonite in- 
crease the number of montmorillonoid 
layers in the structure. The montmoril- 
lonoid materials were found to be 
trioctahedral, indicating that they are 
more closely allied to the biotites, which 

1 1 1 1 
2 mica + qtz + sonidine 


^ — " 



phenqite _— -*■""" 

_^--*^IO 20 30 




^ ^,4.5Kb 

15 40 ^-*"*lbo 100 
mica + mont + sonidine -"" mont + sanidine 


+ qtz -t-qtz 


.[... I l t 

Mole % 

MgAI 5 Fe T 5 J 

Fig. 48. The muscovite-(Mgi. Al .6Fe +3 o.5) 
celadonite compositions that result in various 
mica-montmorillonite mixed layerings. Numbers 
represent the percentage of montmorillonoid 
layers present in the aggregate structure. The 
solid line is the boundary between all-mica and 
mica-montmorillonoid phases at 2 kb. The 
dashed line represents this boundary at 4.5 kb. 

they form at higher temperatures (shown 
in the two-mica area at high tempera- 
tures, fig. 48). Montmorillonoids forming 
from starting materials near the musco- 
vite end of the series were dioctahedral. 
Apparently the montmorillonoid that 
forms is a "defect-type" structure (i.e., 
low lattice charge and low potassium 
content) of the mica that it forms at high 

This information can be applied to the 
problem of determining the nature of the 
illite mineral group. Assuming analogy 
between the mixed-layer sequences of 
montmorillonoid micas and mixed-layer 
illites, it could be deduced that illite is a 
low-temperature mixture of two mineral 
compositions that are stable as separate 
phases at higher temperature. Experi- 
ments with natural illites show that 
temperatures of about 300°C bring an 
"unmixing" of the single-phase mixed- 
layer mineral into a mica + chlorite + 
quartz assemblage. This marks the ap- 
pearance of what could be called a 
greenschist assemblage. Experiments at 2 
kb and 300° to 500°C with natural 
materials plus KOH solution give results 
indicating a prohibition of the unmixing 



process, probably through the formation 
of two micas, phengite + glauconite, from 
the initial illite. The common occurrence 
of mica -f- chlorite in rocks suggests that 
the potassium in the total rock compo- 
sition is generally below that needed to 
form micas or micas plus a montmoril- 
lonoid as in the muscovite-celadonite 
systems that were studied. The potassium 
necessary to exclude chlorite from the 
assemblage and to produce celadonite and 
a montmorillonoid would be roughly 10 
per cent K 2 by weight in a shale. 
Potassium in pelitic sediments is usually 
of the order of 3 or 4 weight per cent, and 
therefore mica + chlorite assemblages 
are found. 

It appears that the following sequence 
of mineral changes would be appropriate 
for pelitic sediments: mixed layer "illite" 
— > chlorite + illite — > chlorite + mica, 
with quartz in excess in all steps. The 
amount of mica formed is probably con- 
trolled by the amount of potassium 
present. Metamorphism under high water 
pressures would result in muscovite with 
appreciable amounts of celadonite in 
solid solution. 

Muscovite-Phengite Micas 

The reaction muscovite + biotite + K 
feldspar + quartz + H 2 — > phengite is 
dependent on P H ,o, because water is a 
constituent of the reaction. A similar 
reaction written for chlorite + K feldspar 
+ quartz reactants also favors a phengite 
product under high water pressure. The 
mineral glaucophane appears to have a 
similar relationship (Ernst, 1963). Field 
associations of these two minerals indicate 
that they are found in rocks that have 
been subjected to high pressures. 

Experimental studies in the muscovite- 
MgAl celadonite series demonstrate that 
occurrence of phengites (intermediate 
compositions) (fig. 49) is significantly 
dependent upon pressure. The surface 
depicting the upper stability of phengitic 
micas has been determined at three 
pressures (Ptotai = Ph s o) : 2, 4.5, and 10 
kb. Reactions were reversed at 2 and 4.5 



Mole per cent 

Fig. 49. Representation of the phengite-2 
mica + sanidine + quartz boundary in a P-T-X 
diagram. The surface delineates the maximum 
solid solution of MgAl celadonite composition 
in a muscovite structure. 

kb. The inverse relationship of pressure 
and temperature in the production of 
solid solutions is of interest. Increased 
solid solution is favored by lower temper- 
ature or higher total water pressure. 
Recalling that the reaction product 
phengite (a solid solution between mica 
and celadonite) is favored by high water 
pressure, it can be said that the stability 
of this phase will be affected by a third 
variable, the partial pressure of water in 
the system. With such conditions imposed 
upon the stability of phengites, their 
occurrence in nature can be attributed to 
a rather specific set of physical conditions, 
namely low temperatures, high total 
pressure, and high water pressures. The 
conditions of P Hs o = Ptotai are probably 
often realized in low-grade metamorphism 
of pelitic sediments. Such sediments 
usually contain considerable amounts of 
connate water and water bound to hydro- 
phyllic clay minerals such as montmoril- 
lonoids or vermiculites. The collapsing of 
hydrated minerals, due to higher total 
pressures (as seen in fig. 48) and general 
compaction of the sediments, provides 
enough water to transmit external pres- 
sures upon the rocks. Thus P to tai = Ph 2 o 
could be easily realized. 

It would seem, however, that high total 



pressures, such as 10 kb, would more 
likely be the result of tectonic stress than 
of hydrostatic pressures resulting exclu- 
sively from load effects. A total pressure 
of 10 kb indicates depth of burial in 
continental situations of about 30 km or 
near the base of the crust, where esti- 
mated geothermal gradients suggest tem- 
peratures of 400° to 600°C. The fact that 
at these conditions most natural phen- 
gites would not be stable appears to 
eliminate deep burial as a probable 

condition of phengite origin. Tectonic 
pressures are rather indefinite because 
little is known about the competence of 
rocks under high-pressure conditions. 
However, it seems reasonable to expect 
tectonic factors to lead to phengite 
formation. Field occurrences of phengite 
in metamorphic areas of Japan, Califor- 
nia, and Alpine regions (Ernst, 1963) 
suggest that the requisite PT conditions 
for phengite formation were primarily of 
tectonic origin. 


A growing amount of geophysical and 
petrological evidence indicates that the 
dominant rock type in the upper mantle 
is an aluminous peridotite. Most of the 
experimental work at high pressures this 
year has been directed at increasing our 
understanding of the mineral equilibria in 
such peridotites and of their melting 
relations. These experiments show that a 
peridotite whose composition approaches 
a mixture of 1 part basalt and 3 parts 
dunite can crystallize in three facies. 
These facies, stable under different PT 
conditions, are spinel peridotite, garnet 
peridotite, and a peridotite in which the 
A1 2 3 is entirely in solution in the 
pyroxene. Probable geothermal gradients 
for oceanic regions and continental shield 
areas indicate that the spinel peridotite 
facies should be present beneath the 
oceans to a depth of about 60 km, but 
that it should be absent or very thin 
beneath the shields. This speculative 
picture is in harmony with the natural 
distribution of spinel and garnet perido- 
tites that have been erupted from the 
mantle into the crust. Increasing pressure 
causes A1 2 3 in solution in pyroxene to be 
exsolved as garnet. Determination of 
pyroxene- garnet phase relations has pro- 
vided a quantitative explanation of a 
seismic discontinuity in the upper mantle 
between 150 and 200 km. These phase 

relations also explain the low A1 2 3 
contents of enstatites from ultramafic 
inclusions in kimberlite. Study of melting 
relations in the system diopside-forster- 
ite-pyrope indicates that the low-melting 
fraction in this complex "synthetic perid- 
otite" system will be poor in forsterite 
and will approach the composition of 

Synthesis of the pyroxene ferrosilite 
and study of the stability and melting 
relations of jadeite supply further, dra- 
matic evidence of the changes produced 
by high lithostatic pressure in silicate 
systems. Ferrosilite, FeSi0 3 , has defied 
synthesis for years, but a stability field 
for it has been discovered at high pres- 
sure. The complex polymorphism exhib- 
ited by this phase provides a challenge to 

Further study of the calcite-aragonite 
equilibrium has disclosed a nonquench- 
able phase change that appears to explain 
the presence of calcite rather than 
aragonite in high-grade metamorphic 

Petrological Constitution 

of the Upper Mantle 

A. E. Ringwood, I. D. MacGregor, and F. R. Boyd 

New experimental data on garnet- 
pyroxene equilibria make it possible to 
interpret some features of the seismic 



velocity distribution in the upper mantle. 
In order to apply the experimental results 
quantitatively, specific assumptions about 
the chemical composition of the mantle 
are necessary. It can be argued plausibly 
that the chemical composition of the 
primary undifferentiated upper mantle 
should be such that it can yield a basalt 
magma on partial melting, leaving behind 
the residual, refractory dunite or perido- 
tite. Accordingly, the primary composi- 
tion of the upper mantle would lie be- 
tween those of basalt and peridotite. A 
chemical and penological model for the 
upper mantle based upon this postulate 
has been developed by Ringwood (1962a, 
b), Green and Ringwood (1963), and 
Clark and Ringwood (1964). In this 
model, the primary undifferentiated com- 
position of the upper mantle is assumed 

equal to approximately 1 part of basalt 
to 3 parts of dunite or peridotite. This 
primary material is called pyrolite. A 
typical composition for pyrolite is given 
in table 9, column 1. Also in table 9 are 
given the analyses of four natural ultra- 
basic rocks closely approaching the 
pyrolite composition. The rocks in col- 
umns 2 and 3 display the mineralogy 
olivine -f- aluminous orthopyroxene + 
aluminous diopside + spinel, and may be 
called pyroxene pyrolite. Those in col- 
umns 4 and 5 are composed of olivine + 
enstatite + diopside (both poor in A1 2 3 ) 
+ pyropic garnet, and may be referred to 
as garnet pyrolite. The data in table 9 
demonstrate the capacity of rocks ap- 
proaching the pyrolite composition to 
crystallize in different mineral assem- 
blages having distinctive physical prop- 

TABLE 9. Chemical Analyses and Mineral Assemblages of Rocks 
Approaching the Pyrolite Composition 








+ aluminous 


-f- low-alumina 


pyroxenes + spinel 


+ pyrope garnet 






Si0 2 
















Fe 2 3 




(Total Fe as FeO) 






A1 2 3 












Na 2 






K 2 





Cr 2 3 

















Ti0 2 











P 2 6 






H 2 






C0 2 








* Recalculated to 100 per cent anhydrous. 

References : 

1, 4. Green and Ringwood, 1963. 

2, 3. Green, 1963. 

5. Milliard, 1959 (average of two analyses of garnet peridotite). 



erties (to be discussed). If rocks ap- 
proaching this composition are wide- 
spread in the mantle, it is clear that 
zoning controlled by the PT stability 
fields of these assemblages must occur and 
that the zoning will have an important 
effect on the distribution of seismic 
velocities and densities in the upper 

The concentration and distribution of 
trivalent elements (particularly alumi- 
num) in pyroxene pyrolite strongly 
influence its stability field with respect 
to garnet pyrolite. In rocks resembling 
pyroxene pyrolite, Al 3 , Cr 3 , and Fe 3 are 
partitioned between enstatite, diopsidic 
clinopyroxene, and spinel. Results de- 
scribed in this report and also unpub- 
lished data of MacGregor show that 
spinel is unstable in rocks of pyrolite 
composition above about 1000°C, owing 
to the solubility of Al 3 , Cr 3 , and Fe 3 in 

Experiments described in this report 
deal chiefly with the solid solubility of 
pyrope garnet in orthopyroxene as a 
function of temperature and pressure. In 
order to discuss stability fields in the 
mantle some information on the corre- 
sponding solubility of pyrope garnet in 
diopsidic clinopyroxene is required. Using 
minerals obtained from a garnet perido- 
tite inclusion in a diamond pipe, Mac- 
Gregor and Ringwood (this report) found 
that the solubility of garnet in clino- 
pyroxene is slightly higher than in 
orthopyroxene at similar temperatures 
and pressures. This result is supported by 
comparison of analyses of coexisting 
orthopyroxenes and clinopyroxenes from 
kimberlite and basalt nodules (Mac- 
Gregor and Ringwood, this report). 
Comparison is made on the basis of 
molecular per cent of R 2 3 (A1 2 3 + 
Fe 2 3 + Cr 2 3 ) after subtracting suffi- 
cient R 2 3 to match the Na 2 in the 
analysis as pyroxene, NaRSi 2 06. When 
this is done it is usually found that nat- 
ural diopsidic clinopyroxene contains 
slightly more mole per cent R 2 3 than 
coexisting enstatite, corresponding to a 
higher content of Active garnet. 

An attempt is made in figure 50 to use 
the experimental data obtained so far to 
outline the mineralogical stability fields 
in ultrabasic rocks approaching the 
pyrolite composition. Because of insuffi- 
cient quantitative data on clinopyroxene- 
garnet solubility, it has been assumed that 
the solubility of pyrope garnet in clino- 
pyroxene is similar to that of pyrope in 
orthopyroxene under similar PT con- 
ditions. As was mentioned above, this is 
not strictly correct. Nevertheless, the 
error introduced by this assumption is 
probably small. The effect of a higher 
solubility of garnet in clinopyroxene 
would be to displace the garnet stability 
field of figure 50 to somewhat higher 

In the model, the maximum average 
R 2 3 content of the pyroxenes is taken 
as 4.5 mole per cent, corresponding to the 
situation where all the R 2 3 in the rock 
occurs in solid solution in pyroxene and 
spinel is not present as a stable phase. 
This is close to the highest values that 
would be obtained for the naturally 
occurring assemblages (table 9, columns 
2, 3, 5). Alumina is, of course, by far the 
most abundant of the trivalent oxides 
under consideration. In figure 50 the 
positions of the pyroxene solid solubility 
limits at 1500°C are taken from the 
experimental values obtained for natural 
pyrope and enstatite (MacGregor and 
Ringwood, this report) and the slopes of 
the R 2 3 contours are taken from experi- 
ments on the synthetic system (Boyd and 
England, this report). 

The position of the pyrolite solidus in 
figure 50 is little more than a plausible 
guess based on miscellaneous results 
including those of O'Hara (Year Book 62, 
pp. 71-76). On figure 50 the region ACB 
corresponds to the stability field of the 
assemblage olivine + 2 pyroxenes (4.5 
mole per cent A1 2 3 ) -f- spinel. Along BC, 
in the system MgO-Si0 2 -Al 2 3 , the solu- 
bility of A1 2 3 in orthopyroxene in 
equilibrium with MgAl 2 4 reaches 4.5 
mole per cent. Hence spinel would be 
finally consumed at this boundary accord- 
ing to the reaction 



Depth km 

100 150 


30 40 50 

Pressure kilobars 

Fig. 50. Mineral stability relationships in ultramafic rocks approaching pyrolite composition, 
and their relationship to possible oceanic and Precambrian shield geotherms. 

zMgAl 2 4 + (1 + z)MgSi0 3 

Spinel Enstatite 

= MgSiOs-zAlaOs + zMg 2 Si0 4 (1) 

Aluminous enstatite Forsterite 

where x = 0.045. 

The line AC is the boundary of the 
reaction (MacGregor, this report) 

4MgSi0 3 + MgAl 2 4 

= Mg3Al 2 Si 3 0i2 

+ M g2 Si0 4 (2) 

In the more complex natural system, solid 
solution effects would broaden this linear 
boundary into a transition zone. To the 
right of AC the phases consist of pyrope 
garnet and pyroxene containing varying 
amounts of alumina plus minor Cr 3 and 
Fe 3 . As pressure increases at constant 
temperature, aluminous pyroxenes con- 
tinue to break down into garnet and less 
aluminous pyroxenes according to the 

3MgSiOs-zAl 2 8 = zMg 3 Al 2 SiOi 2 + 

Aluminous enstatite Pyrope 

3(1 - z)MgSi0 3 (3) 


where x < 0.045. 

Above the boundary BC where spinel 
disappears and along the boundary CD 
garnet is formed with increasing pressure 
primarily by the breakdown of aluminous 
pyroxenes according to reaction 3. The 
slope of the boundary defining the first 
appearance of garnet changes around BC 
because the equilibria by which garnet is 
formed (2 and 3) also change in this 

The light broken lines parallel and to 
the right of CD define the alumina con- 
tent of orthopyroxene in equilibrium with 
garnet. It is seen that the formation of 
garnet from aluminous pyroxenes extends 
continuously over a wide pressure inter- 



Having described the stability fields of 
garnet and aluminous pyroxene assem- 
blages in the pyrolite model, we will 
consider some of the properties of this 
model and their bearing on the mantle 
and the genesis of certain ultrabasic 

Seismic velocities in the upper mantle. 
In many parts of the world seismologists 
have observed a break in the travel-time 
curve for P body waves in the vicinity of 
14 degrees. This break was well defined 
beneath eastern United States and west- 
ern Europe (Lehmann, 1959, 1962) but 
was not found beneath the Canadian 
Precambrian shield (Brune and Dorman, 
1963). Lehmann suggested that the break 
could be caused by a sudden increase in 
P wave velocity from 8.12 to 8.35 km/sec 
occurring at a depth of 220 km. She found 
a corresponding discontinuity using S 
waves (Lehmann, 1961). Although there 
is a large uncertainty in the velocity- 
depth profile obtained from the body 
wave travel times in this interval, the 
general validity of Lehmann's model was 
supported by the surface wave investiga- 
tions of Takeuchi, Saito, and Kobayishi 
(1962). More recently, an elaborate 
surface wave investigation of the struc- 
ture of the suboceanic mantle by Ander- 
son (1964a, b) has also yielded results 
generally consistent with those of Leh- 
mann. Anderson found that an increase 
in velocity of about 3 per cent was re- 
quired between 150 and 200 km. Below 
200 km the velocity gradient decreased, 
remaining small until 400 km. 

It can readily be shown that a 3 per 
cent increase in seismic velocity between 
150 and 200 km cannot possibly be caused 
by normal self-compression acting on the 
minerals of this region, particularly when 
the effect of increasing temperature with 
depth is considered (Birch, 1952). In 
fact, the increase attributable to self- 
compression is smaller than 1 per cent. It 
follows that the mantle is inhomogeneous 
in this region and that the increase in 
seismic velocity is caused by changes 
either of phase or of chemical composition. 

Clark and Ringwood (1964) suggested 
that the velocity increase might be caused 
by the transition from pyroxene pyrolite 
to garnet pyrolite. It is now possible to 
test this idea. 

The seismic P wave velocities for 
pyroxene pyrolite and garnet pyrolite 
have been estimated using available data 
on elastic constants and densities of 
minerals and measured seismic velocities 
in rocks and minerals (Birch, Schairer, 
and Spicer, 1942; Birch, 1960, 1961; 
Verma, 1960). The velocity for pyroxene 
pyrolite is 8.30 km/sec and for garnet 
pyrolite 8.53 km/sec (at NTP). The 
densities of these assemblages are 3.31 
and 3.37 g/cm 3 , respectively. Using 
Birch's (1961) velocity-density relation- 
ship (solution 4), the velocities are found 
to be 8.41 km/sec and 8.60 km/sec. The 
seismic velocity of garnet pyrolite is 
therefore about 23^ per cent greater than 
that of pyroxene pyrolite. This is almost 
exactly the increase in seismic velocity 
required by Lehmann and Anderson. 
(Absolute velocities are not strictly 
comparable, since the calculations and 
observations apply to different tempera- 
tures and pressures. However, the effect 
on the velocity difference between the 
two assemblages is insignificant.) 

A possible geotherm for the suboceanic 
mantle has been placed on figure 50. It 
is somewhat higher than that given by 
Clark and Ringwood (1964) and corre- 
sponds to a higher opacity in the upper 
mantle than that assumed by these 
authors. Along this geotherm pyroxene 
pyrolite is stable to a depth of approxi- 
mately 135 km. At this depth garnet 
appears and the garnet pyrolite field is 
entered. As pressure increases, more 
garnet is formed, and the transition is 
effectively complete by 220 km. Most of 
the increase in seismic velocity occurs in 
the first half of the transition interval. 
The depth interval over which the veloc- 
ity increase occurs along the chosen 
geotherm is broadly consistent with the 
possible range of seismic depth-velocity 
solutions (Anderson, 19646). The degree 



of correspondence between seismic and 
petrological models of the upper mantle 
is most encouraging. 

A second geotherm representing Pre- 
cambrian shield regions has been placed 
on figure 50. It falls well below the oceanic 
geotherm because of the low heat flow 
and strong upward concentration of 
radioactivity characteristic of shield re- 
gions (Clark and Ringwood, 1964). The 
suggested geotherm is somewhat higher 
than that in these authors' model, corre- 
sponding to the higher opacity assumed 
in the present oceanic model. It is seen 
that the geotherm does not cross sharply 
from pyroxene to garnet pyrolite. In fact, 
these geotherms below the M discon- 
tinuity remain almost exclusively in the 
garnet pyrolite field. Accordingly a 
velocity increase such as that previously 
discussed is precluded. Detailed seismic 
investigations on the Canadian shield by 
Brune and Dorman (1963) showed that 
the velocity increase inferred by Lehmann 
and Anderson for ordinary continental 
and oceanic regions, respectively, did not 

Melting of pyrolite. According to the 
model depicted in figure 50 the phases 
stable at the solidus down to a depth of 
150 km are olivine and aluminous 
pyroxenes. It is therefore to be expected 
that equilibria involving these minerals 
will control the formation of magmas in 
the mantle by direct fractional melting of 
pyrolite. The important role played by 
aluminous orthopyroxene in the pedo- 
genesis of basalt has recently been 
stressed by Green and Ringwood (1964). 
For rocks approaching the pyrolite com- 
position it does not appear likely that 
equilibria involving garnet will play an 
important role in the genesis of magmas 
by direct fractional melting at depths less 
than 150 km. 

Ultramafic Rocks 
F. R. Boyd and I. D. MacGregor 

Experimental data on garnet-pyroxene- 
spinel relations are providing an increas- 
ing insight into the origins of a variety of 

ultramafic rocks and their relationship to 
the upper mantle. When the geochemical 
results are combined with estimates of 
the geothermal gradients in continental 
and oceanic areas based on modern 
heat-flow studies, some remarkable con- 
sistencies are found to emerge. 

The reaction 4 enstatite + spinel ^py- 
rope + forsterite (fig. 51) divides ultra- 
mafic rocks into two broad groups: a 
relatively low-pressure spinel-bearing 
type, and a high-pressure garnetiferous 
type. Generalizing on the natural distri- 
bution of these types, it can be said that 
with few exceptions their occurrences are 
mutually exclusive. The spinel peridotites 
are found as large masses in the axial 
regions of tectonic belts. On a more 
restricted scale they are also found as 
members of layered, mafic intrusives and 
as nodules in alkali and melilite basalts. 
The spinel peridotite nodules are found in 
basalts erupted in both oceanic and con- 
tinental environments. The garnet perido- 
tites also form large masses in certain 
high-grade metamorphic terranes such as 
the European Caledonian and Hercynian 
mountain belts. Field workers disagree 
about whether the mineral assemblage in 
such garnet peridotites reflects local 
metamorphic conditions, or whether the 
peridotites were intruded into the crust 
from the mantle and their minerals thus 
reflect conditions at depth in the mantle 
(e.g., O'Hara and Mercy, 1963; Schmitt, 
1964). Garnet peridotites are also found 
as nodules in kimberlites, where they 
occur in association with diamonds. 

If the upper mantle is composed 
primarily of peridotite, there will be a 
layer of spinel peridotite below the M dis- 
continuity and garnet peridotite at 
greater depth. The thickness of the spinel 
peridotite layer depends on the geother- 
mal gradient and the thickness of the 
crust (fig. 52). The data in figures 51 and 
52 show that the spinel peridotite layer 
under continental shields will be only a 
few kilometers thick or it may be absent 
altogether. This supposition is in harmony 
with the nature of the inclusions in 



a - Sapphinne 
i - Sillimanite 
p - Spinel 

100 200 

Depth, ki lometers 



30 40 50 60 70 

Pressure, ki lobars 




Fig. 51. Phase-equilibria data combined with geo therms for oceanic areas (A) and Precambrian 
shields (B). The geotherms are from Ringwood (this report); the pyrope stability boundary is from 
Boyd and England (Year Book 61, p. 110); the curve for the reaction 4 enstatite + spinel *± pyrope 
-f- forsterite is from MacGregor (this report); and the diamond ±± graphite curve is from 
Bundy, Bovenkerk, Strong, and Wentorf (1961). 

kimberlite. In the course of eruption from 
the mantle into the crust, kimberlites 
have picked up fragments of the crustal 
and mantle rocks through which they 
passed. Among the fragments are a wide 
variety of crustal rocks and garnet perido- 
tites, but spinel peridotites are extremely 

Under the oceans the spinel peridotite 
layer may thicken to as much as 50 km 
(fig. 52). In tectonically active continental 
areas where the geothermal gradient is 
steeper than in stable shield areas there 
should also be a significant thickness of 

spinel peridotite. These experimental and 
geophysical data are thus in harmony 
with the natural distribution of spinel 
peridotites. As was noted above, one of 
the principal kinds of occurrence of spinel 
peridotite is as relatively large masses 
intruded into the axes of geosynclinal 
belts. Alpine-type peridotites may there- 
fore be interpreted as solid intrusions 
from the uppermost part of the mantle. 
The spinel peridotite inclusions in basalt 
might be fragments derived from primary 
mantle rocks, either directly from the 
spinel peridotite layer or indirectly by 










Fig. 52. Speculative section of the crust and upper mantle extending from a Precambrian shield 
to an oceanic region. The section illustrates the phase assemblages stable at different depths for an 
upper mantle of pyrolite composition. Isotherms (faint lines, °C) are extrapolated from geothermal 
gradients given in figure 50. The bulk composition under the continents may be modified by igneous 

inversion of garnetiferous material during 
upward transport. Alternatively, they 
might be rocks that formed as crystal 
accumulates from basaltic magma in the 
uppermost part of the mantle and were 
then picked up and incorporated in lava 
of later eruptions. In either event the 
substantial thickness of the spinel perido- 
tite facies under oceanic and tectonically 
active continental areas explains why the 
inclusions are almost always spinel perid- 
otite rather than garnet peridotite. 

The occurrence of spinel peridotites as 
layers in large intrusions such as the 
Stillwater and Bushveld complexes is also 
in harmony with the experimental data. 
These rocks have crystallized at relatively 
shallow depths in the crust. Some of them 
show evidence of a reaction relation 
between olivine and liquid. Boyd, Eng- 
land, and Davis (1964) have shown that 
a pressure of only 2.3 kb is sufficient to 
eliminate the incongruent melting of 
enstatite, and such rocks have therefore 
crystallized well within the spinel perido- 
tite field. 

It is possible that under conditions of 
particularly intensive regional metamor- 
phism garnet peridotites become stable in 
the lower part of the crust. Garnet 
peridotites such as those in Norway might 

thus have formed by intrusion from the 
mantle or by metamorphism of peridotite 
originating in some other way. As is 
discussed below, the A1 2 3 contents of 
the pyroxenes in these rocks may provide 
a clue to their origin. 

Experimental data giving the solubility 
of A1 2 3 in enstatite crystallized in 
equilibrium with pyrope are shown in 
figure 51 as a series of isopleths. These 
isopleths are for the pure system MgSi0 3 - 
Mg3Al 2 Si 3 0i 2 . Natural enstatites and 
pyrope-rich garnets contain small 
amounts of ferric iron and chromium 
which substitute for aluminum in the 
garnet and pyrope structures. Experi- 
mental results obtained with mixtures of 
natural enstatite and pyrope (MacGregor 
and Ringwood, this report) are in rela- 
tively good agreement with those ob- 
tained for the synthetic system. If the 
comparison of synthetic and natural 
systems is made on the basis of weight 
per cent A1 2 3 the agreement is exact; if 
it is made on the basis of mole per cent 
A1 2 3 + Fe 2 3 + Cr 2 3 there is a small 
difference. The interpretation of the 
difference is ambiguous, but it is clear 
that there will be no major discrepancy if 
natural enstatites poor in Fe 2 3 and 
Cr 2 3 are considered in terms of experi- 



mental results for the synthetic system on 
the basis of weight per cent A1 2 3 . 

A number of detailed petrographic and 
chemical studies of minerals from nodules 
in kimberlite have recently been made 
(O'Hara and Mercy, 1963; Nixon, von 
Knorring, and Rooke, 1963; Banno, 
Kushiro, and Matsui, 1963). Enstatites 
from these rocks have been found to 
contain a uniformly low percentage of 
A1 2 3 ; most analyses indicate 1 to 2 per 
cent. This range of values is in good 
agreement with the experimental data. 
Figure 51 shows the intersection of the 
diamond-graphite curve with the family 
of A1 2 3 isopleths. It is immediately 
evident that enstatites which have crys- 
tallized with diamond and pyrope cannot 
contain more than 4 per cent A1 2 3 under 
any temperature conditions possible in 
the mantle. Kimberlites have been erupt- 
ed for the most part in Precambrian 
shield areas, and the geotherm for shields 
(fig. 51) is thus applicable. This geotherm 
falls between the 1 and 2 per cent iso- 
pleths over a broad depth range within 
the diamond stability field. Evidently 
this is the zone from which the primary 
minerals in kimberlites have come. 

These data indicate that the tempera- 
ture range in which the primary minerals 
in kimberlites have formed should be 
1000°-1300°C. Davis (Year Book 62, p. 
103) has shown that the solubility of 
MgSi0 3 in diopside is essentially the same 
at 30 kb as at atmospheric pressure (Boyd 
and Schairer, 1964). In principle it should 
be possible to use this solvus curve as a 
geothermometer that is independent of 
pressure. Solid solution of MgSi0 3 in most 
diopsidic pyroxenes from kimberlites 
indicates rather low temperatures, 900°- 
1000°C. There are a few analyses which 
plot over 1000°C, and one subcalcic 
chrome diopside described by Nixon, von 
Knorring, and Rooke (1963, p. 1113, anal- 
ysis E3) gives a temperature of 1300°C, 
which is more in the expected range. With 
this principal exception, however, the 
temperatures indicated by the diopsidic 
pyroxenes seem improbably low. 

It is interesting and possibly significant 
that a large majority of diopsidic py- 
roxenes from coarse-grained plutonic and 
ultramafic rocks have compositions indi- 
cating temperatures in the range 900°- 
1000°C. This generality includes clino- 
pyroxenes from such divergent rocks as 
kimberlites, ultramafic nodules from ba- 
salts, and pyroxenes that have crystal- 
lized from slowly cooled, layered intru- 
sives. This may be a coincidence, and 
possibly the pyroxenes in these rocks 
actually have formed in this temperature 
range. But it might also indicate that 
exsolution in slowly cooled pyroxenes 
takes place in two steps and that 1000°C 
is the temperature range during cooling 
at which exsolution to separate grains 
ceases and exsolution lamellae begin to 
form. In the separation of pyroxenes for 
chemical analysis, exsolution lamellae are 
included in a host grain ; but any separate 
grains of enstatite that had exsolved from 
clinopyroxene, or vice versa, would be 
removed in the purification process. If so, 
the temperature estimated from the 
chemical analysis would be too low. The 
long time required to produce exsolution 
textures in pyroxenes would almost 
certainly prohibit a laboratory test of 
this suggestion, but careful petrographic 
study of pyroxene textural relations in 
various rocks might make it possible to 
confirm or dismiss the idea. 

Hypersthenes from garnet granulites 
frequently have much higher A1 2 3 con- 
tents than enstatites from kimberlites. 
Such pyroxenes contain up to 8 to 9 per 
cent A1 2 3 (Eskola, 1952, p. 152). Appli- 
cation of the phase relations in figure 52 
to such rocks is uncertain because 
pyroxenes and garnets from granulites 
generally contain a large amount of iron. 
The effect of iron on the stability of 
garnet is to enlarge its stability field to 
lower pressures; it is not yet known how 
iron will affect the solubility of A1 2 3 in 
enstatite. The A1 2 3 isopleths in figure 52 
slope toward lower pressures at lower 
temperatures, and they indicate that 
A1 2 3 concentrations as high as 8 per cent 



would be possible at PT conditions of 
approximately 700°C at 10 kb. These 
conditions are not improbable for granu- 
lite zone metamorphism, but the extrap- 
olations involved are so wide that such 
estimates cannot be made with confi- 

The larger masses of garnet peridotite 
like those in Norway are mineralogically 
similar to the kimberlite nodules except 
that they do not contain diamond. The 
AI2O3 contents of the enstatites in these 
rocks are low, generally of the order of 
1 per cent (O'Hara and Mercy, 1963). 

The compositions of these enstatites 
could be interpreted as due to relatively 
low temperature of formation (500° to 
600°C) under metamorphic conditions in 
the crust. But if the garnet-enstatite 
assemblage in these rocks is metamorphic 
in origin it would be expected that some 
of the enstatites would have A1 2 3 con- 
tents as high as the values reported for 
hypersthenes in metamorphic granulites. 
An alternative origin for these rocks is 
that they are solid intrusions from the 
mantle, possibly coming from depths less 
than those required to form diamond. 


OJ 800 









Pressure ,(KB) 

Fig. 53. The solid and liquid equilibrium relations for the molar composition 1 forsterite + 1 
pyrope as a function of temperature and pressure. 



The Reaction 4 Enstatite + Spinel 

<=i Forsterite + Pyrope 

I. D. MacGregor 

The reaction 4 enstatite + spinel <± 
forsterite + pyrope helps to define the 
boundary between the stability fields of 
spinel- and garnet-bearing peridotites in 
the lower crust and upper mantle. Four 
crystalline assemblages were used to 
establish equilibrium relations: (1) 4 
enstatite + spinel, (2) forsterite -f- py- 
rope, (3) 90% (4 enstatite + spinel) + 
10% (forsterite + pyrope), and (4) 90% 
(forsterite + pyrope) + 10% (4 enstatite 
+ spinel). 

In runs close to the equilibrium curve 
reaction rates were slow, and at tempera- 
tures below 1300°C little or no reaction 
took place in runs up to 48 hours in 
duration. The addition of water to runs 
below 1300°C resulted in a considerable 
increase of reaction rates and allowed the 
study to be extended down to 1000°C. 

Using starting material 1, no unique 
subsolidus equilibrium curve could be 
established; in general this assemblage 
could be converted to pyrope + forsterite 
only at pressures up to 6 kb in excess of 
the equilibrium pressure for any given 
temperature. Correspondingly, starting 
material 2 gave similar problems, and 
pressures up to 2 kb less than equilibrium 
pressures were required for its conversion 
to 4 enstatite + spinel. The sluggishness 
was interpreted to result from the failure 
of pyrope and enstatite to nucleate within 
certain limits of the boundary curve. 
Thus both starting materials, 1 and 2, 
were seeded with 10 weight per cent of 
the alternative starting material, and a 
more reliable equilibrium curve, falling 
between the above two extremes, was 
established. This curve was reversed at 
1200°, 1400°, and 1600°C. 

The solidus and liquidus relationships 
were outlined using crystalline enstatite 
+ spinel (assemblage 1) as a starting 
material, and in a few runs in the pyrope 
forsterite field crystalline pyrope + 
forsterite (assemblage 2) was used. 

The experimentally determined equi- 
librium boundaries for the subsolidus, 
solidus, and liquidus reactions are given 
in figure 53. Equations for the curves for 
the different reactions were fitted by eye 
and are given as follows: 

1. Subsolidus reaction: 4 enstatite + 
spinel *± forsterite + pyrope. T = 
0.0546P - 173. 

2. Solidus reaction: 4 enstatite + spi- 
nel +± liquid + forsterite + spinel. T = 
0.00569P + 1497. 

3. Solidus reaction: liquid + forster- 
ite + spinel ^=± liquid + forsterite. T = 
0.00533P + 1560. 

4. Liquidus reaction: liquid + forster- 
ite <=± liquid. T = 0.00433P + 1620. 

T is given in degrees Centigrade and P 
in bars. The solidus curves are valid only 
for pressures above 15 kb. 

Linear extrapolation of the liquidus 
curve to 1 atm is in agreement with the 
inferred melting temperature for this 
composition (Osborn and Muan, 1960). 
However, both solidus curves indicate 
that below 15 kb either considerable 
curvature or inflection of the curves is 
necessary for agreement between the 
high-pressure and 1-atm data. 

The System Enstatite-Pyrope 
F. R. Boyd and J. L. England 

The system MgSi0 3 -Mg3Al 2 Si 3 ]2 is 
binary below the solidus, and it consists 
of a wide field of solid solution of alumi- 
nous enstatite separated by a miscibility 
gap from pure pyrope. Pyrope lies on the 
join MgSi0 3 -Al 2 3 , so that the compo- 
sition of an aluminous enstatite can be 
expressed either as an enstatite with a 
particular percentage of A1 2 3 or as a 
member of an enstatite-pyrope solid 
solution series. The most aluminous 
enstatites that can form under equilib- 
rium conditions in this system contain 
about 16 weight per cent A1 2 3 or about 
63 per cent pyrope. 

The alumina content of an enstatite in 
equilibrium with pyrope is a function of 



Weight per cent 

Mg S i03 10 20 30 40 50 60 70 80 90 Mq^ Al 2 Si 3 0j 2 

Weight per cent Al 2 3 

20 22 24 


Fig. 54. The system enstatite-pyrope at a pressure of 30 kb. Open points with light outline are 
Al enstatite crystallized from glass. Open points with heavy outline areAl enstatite formed by homog- 
enizing enstatite-pyrope mixtures. Points with horizontal ruling are enstatite-pyrope mixtures 
which failed to homogenize. Points with vertical ruling represent pyrope exsolved from homogeneous 
Al enstatite. Opaque points are all quench crystals. 

both temperature and pressure. The 
temperature dependence of the equilibria 
is shown in a temperature-composition 
section for 30 kb (fig. 54). The solubility 
of AI2O3 in enstatite increases with 
temperature. Over the range determined 
experimentally it increases from a little 
less than 5 per cent at 1100°C to about 

16 per cent at 1650°C, where the bound- 
ary of the miscibility gap intersects the 
melting interval. In the temperature 
range below 1100°C the miscibility bound- 
ary must curve so that it becomes 
asymptotic to the temperature axis; 
above 1100°C it is linear. 

There is no discernible solid solution in 






< 14 



1 1 1 I 

1 i 

1 ' 1 

1 1 1 1 1 















d nan 




! 1 1 i 


1 1 1 




30 40 50 

Pressure^ki lobars 


Fig. 55. Isothermal sections of the boundary of the miscibility gap in the system enstatite-pyrope. 
The two small circles with crosses at 30 kb are points taken from the isobar in figure 54. Other 
points are quenching data; the circular points apply to the 1600°C isotherm, and the square points 
locate the 1100°C isotherm. The coding of these points is the same as in figure 54. 

pyrope in this system. The cell edge of 
pyrope crystallized in equilibrium with 
aluminous enstatite at 15Q0°C and 35.5 
kb was determined to be 11.455 A, com- 
pared with 11.456 ± 0.002 A determined 
for pure pyrope (Boyd and England, Year 
Book 58, p. 84). The difference is well 
within the precision of measurement. 

The pressure dependence of the equi- 
libria is shown with isotherms at 1100° 
and 1600°C in figure 55. These isotherms 
are projections of the boundary of the 
miscibility gap on the pressure-compo- 
sition plane. They can be read in exactly 
the same way as the more familiar 
temperature-composition section in figure 
54. For each isotherm, runs below the 
curve are in the single-phase field, whereas 
runs above the curve crystallize to 
pyrope + aluminous enstatite. The 
1100°C isotherm is terminated at its low- 
pressure end by the solid state breakdown 
of pyrope, and the 1600° isotherm is 
similarly terminated by the incongruent 
melting of pyrope. Each isotherm passes 
through a point fixed by the 30-kb isobar 
in figure 54; these points are indicated by 

small circles with crosses. The remaining 
data that locate the isotherms are 
quenching runs. 

Increase of pressure greatly reduces the 
solubility of A1 2 3 in enstatite in this 
system, because pyrope is much denser 
than enstatite (3.58 versus 3.20). The 
reaction aluminous enstatite — > enstatite 
(with less A1 2 3 ) + pyrope proceeds with 
a decrease in volume. The effect of pres- 
sure is most pronounced in the range 
below 40 kb. Above 40 kb the isotherms 
curve so that they become asymptotic to 
the pressure axis. Extrapolations of the 
isotherms show that the system enstatite- 
pyrope should have virtually no solid 
solution at pressures above 80 kb. 

Points taken from the two isotherms in 
figure 55 are plotted on a temperature- 
composition projection in figure 56 to give 
a series of isobars. The assumption is 
made that these isobars are linear, as is 
the isobar determined at 30 kb (fig. 54). 
Figure 56 indicates more clearly how the 
miscibility gap between enstatite and 
pyrope widens with increase of pressure. 

The boundary of the miscibility field 






CJ 1400 


a 1300 





I 100 



1 I ' i I I ' I ' I ' I 

I I | I i I I ' 

50 Kb 

MacGregor and Ring wood (1964) 

Al -En + Py 

-/ / 

1 I 1 1 I 

1 1 1 




20 22 24 


Weight per cent Al 2 ; 

Fig. 56. Isobars of the miscibility gap boundary in the system enstatite-pyrope taken from the 
data shown in figures 54 and 55. The small circles are points at 1500°C and 50, 40, and 30 kb taken 
from the data of MacGregor and Ringwood (this report), in which they used mixtures of natural 
enstatite and pyrope. In plotting their data, only the A1 2 3 contents of the mixtures were used; 
small amounts of Fe 2 3 and Cr 2 3 in their samples were neglected. 

can be reversed over a very narrow 
temperature, pressure, or composition 
interval under favorable conditions, but 
in some portions of the phase diagram 
there is a significant hysteresis. If glass is 
used as a starting material, it crystallizes 
first to a homogeneous pyroxene. If the 
run is well inside the two-phase region, 
pyrope slowly exsolves from the super- 
saturated aluminous enstatite. If pyrope 

+ aluminous enstatite are the starting 
materials, they react readily to form a 
homogeneous pyroxene in the single- 
phase field. Fifteen minutes is sufficient 
to make this reaction go to completion at 
1600°C. Exsolution of pyrope from alu- 
minous enstatite proceeds much less 
readily and is somewhat erratic in runs 
close to the boundary of the miscibility 
field. The hysteresis in dry runs in the 




range 1300° to 1600°C at 30 kb is as much 
as 2 per cent A1 2 3 or about 75°. 

The hysteresis seems to be caused by 
the difficulty of nucleating pyrope rather 
than by a sluggish reaction rate. This 
difficulty, first noted in the determination 
of the stability field of pyrope (Boyd and 
England, Year Book 58, p. 85), has since 
been observed in a number of multi- 
component systems in which pyrope is a 
phase. For this reason some preliminary 
values given by us for the solubility of 
pyrope in enstatite in Year Book 62 
(p. 123) were in error. They are corrected 
in the present study. 

At temperatures below about 1400°C 
small amounts of water can be introduced 
into the runs without causing them to 
melt. The H 2 catalyzes the nucleation 
and growth of pyrope, and the hysteresis 
in the miscibility boundary is reduced to 
within the precision of temperature and 
pressure measurement. The boundary of 
the miscibility gap was reversed in the 
30-kb section (fig. 54) between 1250° and 
1275°C at 7.5 weight per cent A1 2 3 . 
Reversals are also shown in the range 20 
to 25 kb for the 1100°C isotherm (fig. 55). 

The usual method of dealing with a 
hysteresis interval in the determination 
of a miscibility boundary is to assume 
that the rates of exsolution or homogeni- 
zation are equal and to draw the bound- 
ary at the midpoint of the interval. In 
the present case it is evident that the rate 
of exsolution in dry runs is controlled by 
sluggish nucleation of pyrope and that the 
rate of homogenization of a pyrope- 
aluminous enstatite assemblage is very 
much faster. For this reason the points 
that locate the miscibility boundary 
above about 1300°C in figures 54 and 55 
were obtained by homogenizing pyrope- 
enstatite mixtures. Below about 1300° 
hydrothermal data obtained by both 
exsolving and homogenizing runs are in- 
cluded. Below about 1100°C the reaction 
rate becomes so slow that a useful 
approach to equilibrium could not be 

The melting relations in the system 

enstatite-pyrope at 30 kb have been 
partly established (fig. 54). There is a 
very narrow melting interval for alumi- 
nous enstatites, and their liquidus inter- 
sects the liquidus for pyrope-rich compo- 
sitions with a minimum at approximately 
en 3B py65. Crystal-liquid equilibria on the 
pyrope side of the system are not binary. 
The field for spinel + liquid is apparently 
restricted to compositions very close to 
pyrope. A similar relationship was found 
by O'Hara {Year Book 62, p. 116) in his 
study of the system diopside-pyrope. 
These data suggest that the incongruent 
melting of pyrope to spinel + liquid will 
probably not be a significant feature of 
magma generation in the mantle, because 
mantle rocks are believed to be pyroxene- 
olivine-garnet assemblages with bulk 
compositions considerably removed from 
pure pyrope. Crystal-liquid equilibria in 
pyrope-rich mixtures other than those 
involving spinel cannot be quenched, and 
it has thus far proved impossible to locate 
the solidus accurately or to distinguish 
primary phases. Pyrope melts congru- 
ently at pressures somewhat above 30 kb, 
and it is probable that at sufficiently high 
pressure this system will become a simple 
binary eutectic. 

The Natural System 


I. D. MacGregor and A. E. Ringwood 

O'Hara and Mercy (1963) have shown 
that the A1 2 3 content of enstatites from 
garnet peridotites is low (1 to 2 weight 
per cent). Garnet peridotite consists of 
the three-phase assemblage forsterite + 
enstatite + pyrope, and the solubility of 
A1 2 3 in enstatite is governed by the 

Al-rich enstatite 

*± Al-poor enstatite + pyrope 

Since the assemblage Al-poor enstatite + 
pyrope has a smaller molar volume than 
Al-rich enstatite, the A1 2 3 content of 
enstatites coexisting with pyrope should 
decrease with increasing pressure at 
constant temperature. Recent work 









1 1 

1 1 ' 






















\ D 





D \ 

\ B 

















\ D 

\ — 













. 1,1, 

2 4 6 8 10 12 

Molecular per cent, total R 2 3 in mixture 

Fig. 57. The pressure dependence of the reaction Al-rich enstatite <=± Al-poor enstatite + garnet 
for a natural enstatite and garnet at 1500°C. The equilibrium curve of the synthetic reaction has 
been determined by Boyd and England (this report). 

(Boyd and England, this report) on the 
reaction in the synthetic system MgO- 
Al 2 3 -Si0 2 has shown that it does. The 
present study is a determination of the 
same reaction using analyzed enstatite 
and pyrope (table 10) from Bulfontein, 
South Africa. 

The apparatus used in this study was 
the single-stage apparatus described by 

Boyd and England (1960, 1963). Starting 
mixtures were composed of different 
ratios of analyzed enstatite and pyrope. 
The products were examined optically, 
and the boundary was drawn between 
runs having the two-phase assemblage 
enstatite + pyrope and runs with only 
enstatite. All runs were made at 1500°C, 
at which temperature complete reaction 



TABLE 10. Chemical Analyses of Enstatite and 

Garnet from a Garnet Peridotite Nodule from the 

Bulfontein Pipe, South Africa 








Si0 2 




Ti0 2 




A1 2 3 




Fe 2 3 




















Na 2 




K 2 




P 2 5 




H 2 0+ 




H 2 0~ 




Cr 2 3 







C0 2 








Density 3.19. 

BLT-1 Garnet peridotite, Bulfontein Pipe, 
South Africa. 

DTP-2 Garnet peridotite, Duitoitspan Pipe, 
South Africa. 

Analyst: A. J. Easton. 

Samples kindly supplied by D. H. Green and 
A. J. Easton. 

was possible within half an hour. The 
experimental results are plotted in figure 

In comparing the natural experiment 
with its synthetic counterpart (Boyd and 
England, this report) there is the problem 
of whether weight per cent A1 2 3 or 
molecular per cent (total R 2 3 less the 
R 2 3 associated with Na and K) should 
be taken as the basis. With the weight 
per cent A1 2 3 as the basis for comparison, 
the orthopyroxene in equilibrium with 
garnet at a specific temperature and 
pressure has essentially the same A1 2 3 
content in both the synthetic and natural 
experiment (Boyd and England, this 
report, fig. 56). Thus it would appear that 
the synthetic experiment should give a 
good approximation of the conditions of 
formation of an enstatite, in a garnet 
peridotite, if the A1 2 3 content of the 

enstatite is known. With the molecular 
per cent (total R 2 3 less the R 2 3 associ- 
ated with Na and K) as the basis, figure 
57 shows that, at 1500°C, the natural 
orthopyroxene at equilibrium at low 
pressures has considerably more R 2 3 
than a synthetic enstatite at equilibrium 
under the same conditions. However, with 
increasing pressure the difference de- 
creases so that at pressures in excess of 
50 kb it is negligible, and both synthetic 
and natural enstatites may be directly 

Preliminary runs on the reaction 

Al-rich diopside 

<=± Al-poor diopside + garnet 

with natural, analyzed mineral pairs 
(table 10) shows that at 1500°C and at 
pressures between 30 and 35 kb the 
diopside on the equilibrium curve has 
approximately 1 molecular per cent more 
total R 2 3 (less the R 2 3 associated with 
Na and K) than the enstatite under the 
same conditions. This agrees fairly well 
with analyses of coexisting enstatite and 
diopside pairs from garnet peridotites 
(O'Hara and Mercy, 1963). 

Aluminous Enstatites 
Brian J. Skinner 10 and F. R. Boyd 

The coupled substitution of Al-Al for 
Mg-Si in the orthorhombic enstatite 
(MgSi0 3 ) structure has been shown to be 
extensive in enstatites synthesized at high 
pressures (Boyd and England, this report, 
p. 157). The same coupled substitution in 
the monoclinic diopside (CaMgSi 2 6 ) 
structure, yielding the Ca-Tschermak's 
molecule (CaAlfAljSiJOe), has been dem- 
onstrated by Clark, Schairer, and de 
Neufville (Year Book 61, pp. 59-68). 

The substitution of Al for the larger 
Mg ion in an octahedrally coordinated 
site should reduce the unit cell volume, 
while the substitution of Al for a smaller 
Si ion in a tetrahedrally coordinated site 
should have the opposite effect — a tend- 
ency for the cell volume to increase. 

10 U. S. Geological Survey, Washington, D. C. 



A series of homogeneous aluminous 
enstatites were prepared at regular inter- 
vals out to 15 weight per cent A1 2 3 and 
examined by X-ray powder diffraction 
techniques. The measurements were in- 
ternally calibrated with an NaF standard 
whose cell edge had in turn been carefully 

calibrated against gem diamond. The 
powder diffraction data were refined and 
indexed, using the least squares computer 
indexing program written for the Bur- 
rough's 220 computer by D. E. Apple- 
man, H. T. Evans, Jr., and D. Hand- 
werker. The cell edges, cell volumes, and 

Weight per cent Al 2 3 

6 8 10 

4 6 8 10 

Weight per cent Al 2 3 

Fig. 58. Unit cell parameters of aluminous enstatites. 



densities calculated for the 14 samples 
examined are plotted in figure 58. The 
internal precision of measurement is 
excellent, as is the agreement between 
samples of the same composition. 

The effect of the Al-Al substitution for 
Mg-Si in enstatite is almost identical with 
that found for diopside (Year Book 61, 
p. 63) ; the a and b axes decrease, reflecting 
the effect of the octahedral site substi- 
tution, whereas the c axis remains 
essentially constant over the composition 
range studied, reflecting the counteracting 
tendency for the tetrahedral site substi- 
tution to cause an expansion. The c-axis 
direction is, of course, the one parallel to 
the chains of (Si,Al)0 4 tetrahedra. 

The cell volume decreases with increas- 
ing alumina content, again similar to 
diopside. Because of the nearly identical 
molecular weights for enstatite (100.41) 
and alumina (101.96), the density in- 
creases with increasing alumina content 
from 3.20 g/cm 3 for pure enstatite to 3.26 
g/cm 3 for an enstatite containing 15 
weight per cent A1 2 3 . Projecting the 
observed density versus composition rela- 
tionship to 25 mole per cent A1 2 3 , the 
composition of pyrope, we obtain a den- 
sity of 3.28 ± 0.01 g/cm 3 for Mg 3 Al 2 Si 3 0i2 
with the orthorhombic enstatite struc- 
ture. This density is considerably less 
than the 3.58 g/cm 3 observed for the 
more densely packed pyrope structure 
and indicates that increasing pressure will 
favor the assemblage pyrope + enstatite 
over aluminous enstatite, in agreement 
with the phase relations found by Boyd 
and England for the system enstatite- 
pyrope (p. 157). 

The System Diopside-Forsterite-Pyrope 

at 40 Kilobars 

B. T. C. Davis 

A wide range of mafic and ultramafic 
compositions crystallize to assemblages 
containing garnet, olivine, and one or two 
pyroxenes at pressures between 25 and 
50 kb, a pressure interval in which many 
basaltic magmas are very probably 
formed. The melting relations of garnet 
peridotite, the ultramafic rock containing 

all these phases, are therefore vital to an 
understanding of the origin of basalts. 
Although fully a dozen oxides are present 
as more than 0.1 per cent in a typical 
analysis of a garnet peridotite (table 11), 
about 90 per cent of any analysis consists 
of Si0 2 , A1 2 3 , CaO, and MgO, so that 
the behavior of this rock type may be 
treated in terms of the four-component 
system if it be assumed that all the other 
components behave like one or another 
of the four major oxides and serve mainly 
to depress the melting interval relative to 
that in the quaternary. This assumption 
has been justified by experiments at 
atmospheric pressure. 

TABLE 11. Typical Analyses of Garnet 
Peridotites from Nodules and Intrusions 





Si0 2 





Ti0 2 








Fe 2 3 





Cr 2 3 























Na 2 





K 2 



H 2 + 





H 2 0~ 





C0 2 



P 2 5 





























1, 2. Large intrusive garnet peridotite with 
associated eclogite, Beni-Bousera, Morocco 
(Milliard, 1959). 

3. No. E-3 lherzolite nodule, kimberlite, 
Thaba Putsoa, Basutoland (Nixon, von Knor- 
ring, and Rooke, 1963, table 1). 

4. No. E-ll lherzolite nodule, kimberlite, 
Lourwencia, South West Africa (Nixon, von 
Knorring, and Rooke, 1963, table 1). 



Within the quaternary CaO-MgO- 
AI2O3-S1O2 the join diopside-forsterite- 
pyrope may provide an insight into the 
behavior of natural garnetiferous perido- 
tite assemblages, for there is reason to 
believe that a large part of this join 
crystallizes to the assemblage garnet + 
two pyroxenes + olivine. Melting rela- 
tions in the three binary and pseu- 
dobinary joins that form the edges 
of diopside-forsterite-pyrope indicate a 
mechanism for the generation of eclogitic 
liquids from garnetiferous peridotites and 
confirm previous indications (O'Hara, 
Year Book 62) that these liquids can 
greatly change in composition by frac- 
tionation at high pressure. Establishment 
of the actual melting relations of gar- 
netiferous peridotites and knowledge of 
the full compositional range of their 
derivative liquids will require further 
studies in the systems diopside-forsterite- 
pyrope and wollastonite-enstatite- 

Of the joins fundamental to diopside- 
forsterite-pyrope, diopside-forsterite has 
previously been examined at 1 atm and 

20 kb (Kushiro, Year Book 62, and this 
report) and diopside-pyrope has been 
studied at 30 kb (O'Hara, Year Book 62). 
However, MacGregor (this report) has 
shown that at 30 kb on the solidus the 
join forsterite-pyrope is unstable relative 
to the join enstatite ss -spinel, so that, 
unless CaO and the iron oxides have a 
major effect on this reaction, the assem- 
blage garnet-two pyroxenes-olivine is not 
stable at this pressure. To avoid compli- 
cations from the presence of spinel in both 
the subsolidus and the melting interval, 
all the limiting joins in diopside-forsterite- 
pyrope have been examined at 40 kb, at 
which pressure the join forsterite-pyrope 
is stable at the solidus. 

The single-stage apparatus of Boyd and 
England (1963) was used in this study, 
with lengths of runs from 5 minutes at 
about 1700°C to 30 minutes at 1650°C. 
The reactants for diopside-forsterite and 
diopside-pyrope were microcrystalline as- 
semblages grown from glass at 1 atm by 
Schairer; mixtures of pure crystalline 
Mg 2 Si0 4 and Mg3Al 2 Si 3 0i2 were used for 
forsterite-pyrope. The compositions diop- 


Mg 2 Si04 


□ Liquid L S Forsterite Fo ■ Garnet G 

Fig. 59. The join forsterite-pyrope at 40 kb 

40 60 

Weight per cent 

S Forsterite Fo 

Mg 3 AI 2 Si 3 |2 



side 5 5-pyrope 45 and diopside 3 5-pyrope 65 
were reexamined using mixtures of the 
pure crystalline end members. All prod- 
ucts were examined both optically and by 

At 40 kb the join pyrope-forsterite is 
binary with a eutectic at 1770°C, 76 
weight per cent pyrope (fig. 59). There is 
no evidence of solid solution on this join, 
for runs of 5 minutes at 1760°C failed to 
homogenize mixtures of 99 weight per 
cent pyrope, 1 per cent forsterite, and 99 
weight per cent forsterite, 1 per cent 
pyrope. In confirmation, the cell edge of 
pyrope crystallized 5 minutes at 1750°C 
in the presence of 10 per cent forsterite 
is 11.457 ± 0.002 A, in good agreement 
with the o value for pure pyrope, 11.456 
± 0.001 A (Boyd and England, Year Book 

The join diopside-forsterite (fig. 60) is 
pseudobinary and nearly identical, except 
for higher temperatures of melting, with 
the same join at 20 kb (Kushiro, this 
report). Subsolidus assemblages at 
1700°C, between 20 and 95 weight per 
cent diopside, lie entirely in the phase field 

diopside a3 + forsterite ss . Only a trace of 
forsterite was observed in the 95 per cent 
diopside run, so that a solubility of nearly 
5 weight per cent forsterite in diopside at 
the solidus can be inferred. This is in 
accord with Kushiro's studies of the same 
system at 1 atm and 20 kb, in which he 
showed that clinopyroxene-olivine tie 
lines are athwart the join diopside- 
forsterite and that at least the pyroxenes 
of very diopsidic composition contain 
appreciable forsterite as well as enstatite 
in solid solution. 

The melting relations too are at least 
qualitatively similar to those observed at 
lower pressures in that the solidus is not 
isothermal and there is a narrow field of 
clinopyroxene + olivine + liquid. Tem- 
perature uncertainties in the single-stage 
apparatus are so great as to make it 
impossible to tell whether there is a 
maximum on the liquidus at the diopside 
end; this is considered a likely possibility 
in view of the 1-atm diagram (Kushiro 
and Schairer, Year Book 62, p. 96, fig. 21). 

A noteworthy feature of the 40-kb 
diagram for diopside-forsterite is that the 











CaMgSi 2 6 

40 60 

Weight per cent 
□ Liquid L 0Clinopyroxene ss Cpx 

£x]Orthopyroxene ss Opx B Garnet ss G 

Fig. 60. The join diopsjde-pyrope at 40 kb. 

Mg 3 AI 2 Si 3 |2 



piercing point di-fo-L at 73 ± 1 weight 
per cent diopside, 1745° ± 10°C, is 
shifted in composition only 4 weight per 
cent toward diopside from its position at 
20 kb (Kushiro, this report), in compari- 
son with a shift of 12 weight per cent 
between 1 atm and 20 kb. The reason the 
change in the piercing point between 20 
and 40 kb is so small is that the diopside 
melting curve flattens from 12.8°/kb at 
1 atm to an average of only 8°/kb be- 
tween 20 and 40 kb, whereas the melting 
curve of forsterite is linear, with a slope 
of 4.8°/kb. Because a change in compo- 
sition of a eutectic or a piercing point 
depends on a difference in slope of the 
melting curves of the components, con- 
vergence of slope of melting curves fixes 
the composition of a eutectic or piercing 
point lying between them. 

The third join, diopside-pyrope, is 
pseudobinary, for it involves both garnet 

and clinopyroxene solid solutions whose 
compositions do not fall on the join, the 
garnets containing grossularite as well as 
pyrope, and the clinopyroxenes bearing 
enstatite and alumina as well as diopside. 
Hence at 1670°C, immediately below the 
solidus, the join passes through five 
different phase fields (fig. 61), two of them 
involving enstatite ss . However, there are 
only two primary phase fields on the 
liquidus, those of clinopyroxene and 
garnet, for orthopyroxene disappears by 
reaction with liquid from all compositions 
in which it is stable at the solidus. The 
join diopside-pyrope is additionally com- 
plicated by peritectic relations, for liquid 
is present at temperatures below the 
invariant solidus of the assemblage garnet 
+ clinopyroxene + orthopyroxene (1690° 
± 10°C). 

Because of these complexities, it is 
helpful to consider diopside-pyrope in 



Fo + Cpx 

L+ Fo + Cpx' 

Forsterite 20 

40 60 

80 Diopside 

Mg 2 Si0 4 

Weight per cent 

CaMgSi 2 6 

□ Liquid L 

S Forsterite ss Fo 

0Clinopyroxene ss Cpx 

Fig. 61. The join diopside-forsterite at 40 kb. 



terms of the ternary join wollastonite- 
enstatite-alumina, which contains the 
compositions of all phases encountered in 
the pseudobinary. A hypothetical liquidus 
and isothermal sections for a part of this 
ternary are shown as figure 62, which has 
been drawn on the basis of data from 
diopside-pyrope, enstatite-pyrope (Boyd 
and England, this report), and enstatite- 
diopside at 30 kb (Davis, Year Book 62). 
Figure 62a shows the general form of 
the liquidus as dark lines, with inferred 
relations at 1670°C, immediately below 
the solidus, projected as dashed lines. The 
four-phase point clinopyroxene-orthopy- 
roxene-garnet-liquid lies outside the three- 
phase triangle for the crystalline phases, 

so that it is a peritectic. The tie line for 
the clinopyroxene-garnet pair in equi- 
librium with orthopyroxene at the solidus 
intersects diopside-pyrope at diopside 4 2- 
pyrope 5 s ± 2, and was drawn by projec- 
tion through this point from the compo- 
sition of the garnet in the three-phase 
assemblage. On the basis of its unit cell 
calculated from (640) and (642) peak 
spacings, this is pyrope 8 6-grossularitei 4 . 

The four-phase point clinopyroxene- 
orthopyroxene-garnet-liquid must lie be- 
tween this line and the join diopside- 
pyrope in order to satisfy the require- 
ments that it be peritectic and that the 
primary phase field of orthopyroxene not 
intersect the join diopside-pyrope. O'Hara 

Hypothetical Liquidus 
and phase fields at 

To Grossularfte 


Fig. 62. Hypothetical melting relations in the system diopside-pyrope-enstatite at 40 kb. Data 
from the join diopside-pyrope; cotectic curves inferred from the joins enstatite-diopside (Davis, 
1964) and enstatite-pyrope (Boyd and England, this report) at 30 kb. Solid solution limits at solidus 
inferred from the above and O'Hara (Year Book 62, fig. 39). a, Projection of the liquidus surface 
(solid lines) on phase fields at the solidus (dashed lines). 6, 1695°C isothermal section, c, 1685°C 
isothermal section, d, 1680°C isothermal section. 



has shown that diopside-pyrope does 
intersect the orthopyroxene primary field 
at 30 kb so that between 30 and 40 kb the 
four-phase point has shifted appreciably 
toward enstatite, making it quite prob- 
able that at 50 kb this point is a eutectic. 
The consequence of such a shift would 
be to prevent most enstatite-bearing 
eclogites and eclogitic liquids from frac- 
tionating toward a residuum with larnite 
in the norm. At 40 kb, however, fractiona- 
tion of a great range of eclogitic liquids 
drives the residual melt away from 
MgSi0 3 , through diopside-pyrope, toward 
the silica-deficient region (fig. Q2b-d). 
Whether residual liquids actually cross 
the join diopside-2 grossular-1 pyrope 
(equivalent to the 1-atm join diopside- 
forsterite-anorthite) and enter the model 
alkali basalt field is not yet known. 

Figure 61 shows that much of the join 
diopside-pyrope crystallizes to garnet + 
two pyroxenes. From figures 59 and 60 it 
is obvious that forsterite has no significant 
effect on the stability of either diopside 
or pyrope. Therefore it seems likely that 
much of the join diopside-forsterite- 
pyrope crystallizes to garnet + olivine + 
two pyroxenes, an assemblage qualita- 
tively representative of garnet peridotite. 

The generation of basaltic (actually 
eclogitic) liquids from such a four-phase 
assemblage devolves basically upon the 
problem of deriving from the four-phase 
assemblage a liquid that will crystallize 
entirely to clinopyroxene -f- garnet. 
Simultaneous disappearance of ortho- 
pyroxene and olivine at a reaction point 
at the beginning of melting is unlikely, 
for it would involve simultaneous appear- 
ance or disappearance of three phases, 
necessitating a degeneracy — all basaltic 
liquids originating by this mechanism 
would have to be formed from unique 
bulk composition, or at a unique pressure 
or temperature, any of which coincidences 
are most improbable. 

Because of the absence of a primary 
phase field of enstatite from the binary 
and pseudobinary joins, it seems likely 
that the liquidus of diopside-forsterite- 

pyrope has only three primary phase 
fields, those of olivine, clinopyroxene, and 
garnet. Because the compositions of none 
of these phases lie on the join, it is almost 
certain that the intersection of their 
primary phase fields will be at a piercing 
point. From figures 59 and 61 the eutectic 
in forsterite-pyrope and the piercing 
point in diopside-forsterite lie quite close 
to pyrope and diopside, respectively. 
Since the piercing point in diopside- 
pyrope is at a considerably lower temper- 
ature than either of the above points, it 
is quite probable that the piercing point 
diopside ss -forsterite ss -pyrope ss -liquid lies 
within a few weight per cent of diopside- 
pyrope. Such a case is shown schemati- 
cally in figure 63, with hypothetical 
cotectic curves as dashed lines. The stabil- 
ity fields at the solidus that are critical 
to the argument are shown by solid lines 
on the assumption that the solubility of 
forsterite in clinopyroxene is unaffected 
by alumina, so that all the phase fields 
present in diopside-pyrope extend into 
diopside-forsterite-pyrope. Field a there- 

Fig. 63. Hypothetical projection of the 
liquidus surface of diopside-forsterite-pyrope 
(dashed lines) and phase fields at the solidus 
upon the plane diopside-forsterite-pyrope. a, 
clinopyroxene ss . b, clinopyroxene ss + garnet S8 . c, 
clinopyroxene 88 + garnet ss + orthopyroxene 88 . 
d-e, saturation limit of forsterite in clinopyroxene 
at the solidus. /, clinopyroxene 8S + olivine 88 . 
g, clinopyroxene 88 + garnet S3 + olivine 8S . h, 
clinopyroxene 88 + orthopyroxene 88 + garnet S3 
+ olivine g8 . 



fore corresponds to the field of clino- 
pyroxene ss in figure 61, field b to the field 
garnet S s + clinopyroxene ss in figure 61, 
and field c to the field clinopyroxene + 
orthopyroxene + garnet in figure 61. The 
line d-e forming a boundary to all these 
fields indicates simply the saturation of 
clinopyroxene with forsterite at the 
solidus of diopside-forsterite-pyrope. 
Fields /, g, and h are thus the stability 
fields clinopyroxene ss + olivine ss at the 
solidus, clinopyroxene ss + garnet ss + 
olivine ss , and clinopyroxene ss + garnet ss 
+ orthopyroxene ss + olivine ss , respec- 

If the piercing point lies within the 
two-phase field garnet ss + clinopyroxene ss 
as shown (fig. 63, field 6), partial fusion 
extracts of model garnet peridotites in 
this system, if separated from their source 
material, would crystallize at high pres- 
sure to garnet and clinopyroxene, or to 
equivalent basaltic compositions at 1 atm. 
The possible high-pressure crystallization 
paths for such liquids would be quite 
similar to those of compositions in the 
join diopside-pyrope. 

The most obvious alternative method 
of explaining the relationship between 
eclogites and garnet peridotites is to 
invoke a primary phase field of olivine in 
the join diopside-pyrope-enstatite allow- 
ing liquids derived from the four-phase 
assemblage to lose olivine and ortho- 
pyroxene by successive passage through 
two reaction points. This hypothesis is 
not compatible with the absence of fields 
containing olivine from the system diop- 
side-enstatite (fig. 61) and must be 

High-Pressure Melting Relations for 

Jadeite Composition 

P. M. Bell 

The system jadeite-diopside provides 
an approximation of the distribution of 
sodium and calcium in the mantle that is 
analogous to the relationship between 
albite and anorthite in the crust. Solid 
reactions for jadeite composition have 
been studied by Robertson, Birch, and 

MacDonald (1957). For the melting rela- 
tions, they predicted a shift of the 
nepheline-albite binary eutectic toward 
nepheline, with the result that albite 
should be stable on the liquidus with 
jadeite melting congruently at high pres- 
sures. Knowledge of the melting of jadeite 
would facilitate the study of the binary 
jadeite-diopside. Information on this 
system could be expected to be relevant 
to understanding the processes of differen- 
tiation in the mantle. 

The present study was intended to 
determine equilibrium and test the possi- 
bility of congruently melting jadeite. The 
results would complement the data for 
albite (NaAlSi 3 8 ), which is also part of 
the nepheline-quartz binary. For this 
reason it was important to follow closely 
the methods, calibrations, and single- 
stage techniques of Boyd and England's 
(1963) albite study. In this system it is 
essential to dry the starting materials 
thoroughly, and therefore it was easier to 
approach most of the experimental deter- 
minations with nepheline + albite. A few 
runs with jadeite were, of course, re- 
quired. The results, given in figure 64, 
confirm the predictions of Robertson, 
Birch, and MacDonald (1957) of a shift 
of the eutectic toward nepheline. Jadeite 
was melted at 34, 36, 39.5, and 43 kb. 
Melting was also reversed by converting 
jadeite glass to jadeite at 43 kb. Sub- 
solidus relations of albite + nepheline = 
2 jadeite were studied as a function of 
pressure at 1000°, 1100°, and 1200°C. 
Reversibility of the reaction was demon- 
strated at these temperatures. The curve 
through the points plotted in figure 64 
fits an extrapolation of a Robertson, 
Birch, and MacDonald (1957) curve 
obtained using crystalline materials. A 
few runs (fig. 65) starting with albite were 
made to compare the present calibrations 
with those of Boyd and England (1963). 
The agreement was satisfactory. 

On the basis of the data of figures 64 
and 65, temperature-composition sections 
have been constructed (fig. 66). This is 
possible since the pressure-temperature 











i r 

i r 






20 25 30 
Pressure, Kb 





Fig. 64. Experimental results for jadeite composition. 

planes for the two compositions form 
nearly mirror image conjugates, with 
most of the important chemical features 
falling into the central composition region 
of the binary. The reason for this can be 
observed in figure 66, as the eutectic 
shifts from the albite side of jadeite to the 
nepheline side. The result is the incon- 
gruent melting of albite + nepheline to 
albite + liquid, leaving a small zone for 
jadeite to melt incongruently. At higher 
pressures jadeite melts congruently. In- 
formation on the behavior of compositions 
between albite and jadeite cannot be 
obtained from figures 64 and 65 but is 
evident from figure 66. In these compo- 
sitions a eutectic develops that is shallow 
enough to permit estimation of its posi- 
tion on the temperature-pressure plane. 
It would plot as a continuous curve 
connecting the two sets of invariant and 

singular points, along which the liquid 
changes composition. Accompanying this 
eutectic shift is the rapid increase in the 
melting temperature of jadeite with 
pressure. Albite finally becomes unstable, 
and the melting point at a specific pres- 
sure depends on the composition along 
the jadeite-quartz eutectic. 

The consequences of these results can 
be observed in figure 67, where the 
present curve for congruent jadeite melt- 
ing is plotted with the diopside melting 
curve of Boyd and England (1963). The 
similarity of the shapes of the two curves 
is patent, but the jadeite curve lies at 
temperatures considerably below those of 
the diopside curve. For instance, at 30 kb 
diopside melts 300°C higher than jadeite. 
The effect of these differences in melting 
is to tend to concentrate sodium and 
calcium into the liquid and solid, respec- 








20 25 30 35 

Pressure, Kb 




Fig. 65. Experimental results for albite composition. The albite melting curve is after Boyd and 
England (1963). Dashed curves are based on unpublished data supplied by Boyd. 



I bar 

\\ Carnegiete 

-*-r~ rCarnegiete 
_ Ne |Ne\ + Liq U i d / 



Jd Ab 



L+ \ d \Jd 




+Ab / 

Ab\ / L+Q _ 


Ab + Q 

Jd + Q 


Jd Ab 

10 kb 

Ne + Ab 



Jd Ab 

Jd+U 30kb 

L+Jd> J ,L+A / 

^vf^T^V L+Q 

' " \ I — ^ 

Ne+Jd x Jd+Ab 

A + Q 

Jd + Q 


Jd Ab 





35 kb 

Jd+LN^/ L+Q 

Jd + Q 

Q Ne 

Jd Ab 

Fig. 66. Temperature-composition diagram of various pressures. Inferred boundaries are dashed. 
The 1-bar diagram is inferred after Greig and Barth (1938). 



20 30 40 

Pressure . Kb 

Fig. 67. Jadeite and diopside melting curves. 
The diopside curve is after Boyd and England 

tively. A similar fractionation in 
mantle might be expected. 


Synthesis and Stability of Ferrosilite 

D. H. Lindsley, I. D. MacGregor, 

and B. T. C. Davis 

Although FeSi0 3 is an important com- 
ponent of many pyroxenes, the pure end 
member ferrosilite has never been posi- 
tively identified as a mineral. Bowen 
(1935) found in the lithophysae of several 
obsidians small needles of pyroxene whose 
optical properties closely approach those 
predicted for clinoferrosilite. Chemical 
data are lacking, however, and Bowen 
suggested the possibility that appreciable 
amounts of MnO might be present. 
Furthermore, the observations of Bowen 
and Schairer (1932) that fayalite and 
Si0 2 form from FeSi0 3 bulk compositions 
in controlled atmosphere and evacuated 
tube experiments have been substantiated 
by numerous later workers. After a 
review of the "ferrosilite problem" (Year 
Book 62, pp. 91-92) Yoder, Tilley, and 
Schairer conclude: "At present it may be 
assumed that the compound FeSi0 3 does 
exist. The possible polymorphism of 
FeSi0 3 introduces further complications 

into the meaning of the inversion curve 
of Bowen and Schairer." 

In the past year we have synthesized 
ferrosilite under high pressures and tem- 
peratures in solid-media piston-and-cylin- 
der pressure apparatus. Runs were made 
in iron capsules with tightly fitting lids. 
Because the charge is in equilibrium with 
metallic iron the Fe 2 3 content is prob- 
ably low. (Bowen and Schairer found 
about 2 per cent Fe 2 3 in liquids near 
FeSi0 3 composition and considerably less 
below the solidus.) Addition of oxygen 
from outside the capsule should, by 
reaction with the capsule, increase the 
FeO content of the charge and result in 
the formation of fayalite. As no excess 
fayalite is observed, it appears that the 
capsules are effectively closed to oxygen. 
It is possible that the partial pressure of 
oxygen outside the capsule is buffered at 
low values by reaction with the graphite 
furnace used in the apparatus. 

Preliminary work (Lindsley, Davis, 
and MacGregor, 1964), based on the 
synthesis of ferrosilite from equimolar 
mixtures of fayalite and Si0 2 , indicated 
the existence of two FeSi0 3 polymorphs — ■ 
orthorhombic and monoclinic. It ap- 
peared from the synthesis data that 
orthoferrosilite was the high-temperature 
and clinoferrosilite the low-temperature 
form. More recent work, however, re- 
ported below, shows that each polymorph 
had been synthesized outside its stability 
field. Through the courtesy of Boyd and 
England we have been able to synthesize 
relatively large volumes of ferrosilite as 
starting materials for equilibrium experi- 
ments, the results of which are summa- 
rized in figure 68. All phase boundaries 
involving ferrosilite have been reversed 
except for those between ferrosilite poly- 
morphs and liquid + coesite. 

Items of interest in figure 68 are: the 
existence of three FeSi0 3 polymorphs, 
with orthoferrosilite as the low-tempera- 
ture, high-pressure polymorph; incongru- 
ent melting of ferrosilite in the pressure 
range investigated, with a negative olT/dP 
of the solidus from about 41 to 46.5 kb; 



Pressure , kb 

Fig. 68. Projection onto PT plane of phases and assemblages of FeSi0 3 bulk composition, showing 
stability relations of ferrosilite. The slight refractions of curves shown at several triple points 
were not detected experimentally, but have been drawn in accord with the theoretical distri- 
bution of univariant curves about an invariant point. CI Fs, clinoferrosilite; Coes, coesite; Crist, 
cristobalite; Fay, fayalite; Fs III, ferrosilite III; L, liquid; O Fs, orthoferrosilite; Q, quartz; Tridy, 
tridymite. Fields involving tridymite and cristobalite are approximate only. 

and an indication that FeSi0 3 is not 
stable at any temperature at atmospheric 

The three polymorphs of FeSi0 3 
quenched from high pressure and ob- 
served at 1 atm are: orthorhombic, with 
space group Pbca, corresponding to ortho- 
rather than protoenstatite; monoclinic, 
here called clinoferrosilite, with the 
pigeonite space group P2i/c; and a form, 
tentatively called ferrosilite III, distinct 
from clinoferrosilite but with essentially 
monoclinic optical properties. The struc- 
ture of this third polymorph is not yet 
clear. Burnham is currently making 
single-crystal studies of the three forms. 

Optical properties for each of the three 
polymorphs are given in table 12. The 
optic angle of clinoferrosilite is distinctly 
lower than that predicted by Bowen and 
Schairer (1932). 

Ferrosilite melts incongruently to 
Fe 2 Si0 4 -rich liquid + quartz from 17.5 to 
41 kb and to liquid + coesite from 41 to 
at least 46.5 kb (fig. 68). Both the quartz 
and coesite form well developed crystals 
about 50 microns in greatest dimension; 
in quartz the prismatic form and in 
coesite pseudohexagonal tablets predomi- 
nate. The liquid quenches rarely to glass 
and more commonly to fibrous aggregates 
of fayalite or fayalite + clinoferrosilite. 



TABLE 12. 

Optical Properties of Ferrosilite Polymorphs 


n x 

n Y 





Ferrosilite III 

1.772 ± 0.003 
1.764 ± 0.002 
1.763 ± 0.002 

1.780 ±0.002 
1.767 ± 0.002 
1.766 ±0.002 

1.789 ± 0.002 
1.792 ±0.002 
1.785 ± 0.002 

58° ± 5° 
23° ± 2° 
39° ± 3° 

31° ± 1° 

48° ± 6° 



The negative slope of the solidus bound- 
ary between clinoferrosilite and liquid + 
coesite suggests that that assemblage is 
denser than clinoferrosilite. Clinoferro- 
silite, in turn, is denser than fayalite + 
coesite (table 13), and thus it appears 
that the liquid is denser than fayalite. 
(It should be remembered that although 
the liquid lies near fayalite composition 
it must lie between fayalite and ferro- 
silite. Hence rigorous comparison between 
the liquid and fayalite is not valid.) 

A straight-line extrapolation of the 
lower pressure stability curve for ferro- 
silite (2 ferrosilite = quartz -f- fayalite; 
fig. 68) intersects the 0°C temperature 
axis at roughly 10 kb. Unless the curve 
is strongly refracted by the appearance 
of an undiscovered FeSi0 3 polymorph, or 
has an unusual concave upward curva- 
ture, the assemblage fayalite + quartz is 
stable down to room temperature at 
atmospheric pressure. Thus the high- 
pressure data explain the occurrence of 
the assemblage fayalite + Si0 2 in low- 
pressure experiments, and they strongly 

TABLE 13. Molar Volumes of Assemblages 
with 2FeSi0 3 Bulk Composition 


Molar Volume, 
cm 3 /mole 

Fayalite (46.41) + quartz (22.68) 69.09 

Fayalite + coesite (20.75) 67.16 

2 Clinoferrosilite (33.07) 66. 14 

2 Orthoferrosilite (32.94) 65.88 

Fe 2 Si0 4 spinel (42.02) + coesite 62.77 

Fe 2 Si0 4 spinel + stishovite (14.02) 56 . 04 

Data for ferrosilite polymorphs based on pre- 
liminary unit-cell volumes. References for other 
data given in Lindsley, Davis, and MacGregor, 

suggest that the "ferrosilite" of Bowen 
(1935) is not pure FeSi0 3 . 

The experimental data indicate that 
pure FeSi0 3 is stable only at pressures 
corresponding to the lower crust and 
mantle, where rocks are probably rich in 
MgO. Thus ferrosilite is unlikely to be an 
important mineral petrologically ; how- 
ever, the data from this investigation will 
be useful in optical, crystallographic, and 
thermodynamic studies of the pyroxene 
system as a whole. 

High-Pressure Differential Thermal 

Analysis of a Fast Reaction with 


P. M. Bell and J. L. England 

Calcite is the predominant form of 
CaCOs in nature. Aragonite is relatively 
rare, but it sometimes occurs as a meta- 
morphic mineral. For example, Coleman 
and Lee (1961) found aragonite to be 
widespread in the glaucophane-lawsonite 
schists in the Franciscan series of Cali- 
fornia. Aragonite in these rocks occurs 
with typical members of the glaucophane 
schist facies such as glaucophane, law- 
sonite, pumpellyite, and stilpnomelane. 
The rarity of aragonite implies a rather 
limited field of stability; its purity as a 
natural mineral should permit quanti- 
tative application of the calcite-aragonite 
stability curve of these rocks. The equa- 
tion for this curve is given by P = 5500 
(±500) + 12.6 (±1.5) T (P in bars, T 
in degrees Centigrade), according to the 
oscillating-squeezer experiments of Sim- 
mons and Bell (1963). Aragonite is stable 
on the high-pressure side of this curve and 
should be stable, therefore, in high-grade 
metamorphic rocks. Instead, calcite is 
found. The present study is an attempt 



to extend the work of Simmons and Bell 
to higher temperatures and perhaps gain 
more information on the problem of the 
stability of aragonite and therefore on the 
conditions of formation of the aragonite- 
bearing Franciscan rocks, which might be 
generalized to rocks in other areas of the 
same metamorphic grade. 

In the course of this study it was 
necessary to develop a differential ther- 
mal analysis technique that could be 
operated in high-pressure solid-media 
apparatus. This technique, which employs 
standard principles of 1 atm DTA, 
potentially could be used in rather general 
high-pressure applications of experimental 

Experiments were run with two solid- 
media presses, the Boyd-England single- 
stage, and a Hall-type tetrahedral press, 

which was used at the Solid State Physics 
Laboratory of the Cambridge Research 
Center (Air Force). Results of quenching 
runs with these two presses (fig. 69) agree 
and form an apparent curve quite similar 
to the one determined by Sclar, Carrison, 
and Schwartz (1962) with their Hall belt, 
also a solid-media press. This curve 
diverges widely, however, from a linear 
extrapolation of the Simmons-Bell curve. 
Starting materials for these runs were a 
Baker calcite precipitate, a Merke calcite- 
aragonite precipitated mixture, and arag- 
onite synthesized from the Baker calcite. 
The two materials gave slightly different 
results at the boundaries, but all charges 
in the zone between curves A and B of 
figure 69 appeared to have undergone 
inversion during the quench. The high- 
pressure DTA apparatus was designed to 




3 800 

1 700 

H 600 



18 20 22 24 26 
Pressure , Kb 

30 32 


Fig. 69. Results for CaC0 3 with solid-media pressure apparatus. Single-stage piston and cylinder 
press: rectangles and curve C. Tetrahedral-anvil press: circles and curve C. Oscillating-squeezer 
press: curve A (after Simmons and Bell, 1963). Belt-press: curve B (after Sclar, Carrison, and 
Schwartz, 1962). Open and solid symbols signify calcite and aragonite synthesis, respectively. 
Curve D is Bridgman's (1939) calcite I-II equilibrium detected with his volume apparatus. 



investigate the possibility of a rapid 
reaction. This design is similar to that 
described by Kennedy and Newton 
(1963). The differential couples used were 
chromel-constantan-chromel, chromel- 
alumel-chromel, and platinum-platinum 
10 per cent rhodium-platinum. Tempera- 
ture was plotted against differential 
temperature with a Moseley model 135 
X-Y recorder. Preamplification, when 
required, was gained from a Leeds and 
Northrup linear output null detector. 

The DTA results define a curve within 
the uncertainties, which lies on the 
Simmons-Bell extrapolation (fig. 70). The 
reaction is fast when approached from 
either side of the boundary, and reversi- 
bility is easily demonstrated at each 
point. Unfortunately, the high-pressure 
assemblage that is stable in the region 
between the two curves cannot be 
quenched. There is a small possibility 
that this is an entirely new phase, but it 
cannot be verified at present. Moreover, 
Bridgman (1939) discovered two new 
calcite phases at high pressures. The one 
close to the present experimental range is 
indicated by curve D on figure 69, but no 

relationship to the present problem, stable 
or metastable, seems probable. The fast 
reaction more likely is calcite-aragonite 
with a much increased reaction rate due 
to the higher temperatures involved. If 
so, the region between the two curves is 
a zone from which aragonite cannot be 
quenched. This possibility is supported by 
the textures of the charges run in this 

These experiments provide an explana- 
tion for the lack of aragonite in high-grade 
metamorphic rocks. Curve D (fig. 70) for 
the kyanite-sillimanite equilibrium im- 
plies that aragonite is stable in the 
kyanite zone, but the reason that calcite 
is found in such rocks now becomes 
evident. The explanation is that aragonite 
cannot be quenched to atmospheric pres- 
sure from the zone between curves B and 
C without inverting to calcite. The data 
(fig. 70) also suggest that aragonite- 
bearing rocks that formed at pressures 
less than about 10 kb could not have been 
heated much above 200° to 300°C. In 
general, rocks containing aragonite have 
probably formed at low temperatures and 
relatively high pressure. 

O 1300 
J 1200 


E lioo 


E 1000 


16 18 20 22 24 26 28 30 32 
Pressure , Kb 

34 36 38 40 

Fig. 70. Differential thermal analysis of calcite. Symbols show reversed equilibrium points for 
calcite starting material. Curves A, B, and C are the same as those referred to in figure 69. Curve 
D is for kyanite-sillimanite equilibrium, after Clark (1961) and Bell (1963). 




Petrographic research conducted dur- 
ing the year is reported here in two major 
parts, the first dealing with dominantly 
statistical studies, the second with prob- 
lems in which the emphasis is primarily 
on more conventional aspects of chemical 
and mineralogical petrology. Part I is 
concerned entirely with the application of 
discriminant functions to petrographic 
taxonomy. It opens with a description of 
the two-group discriminant as applied to 
the problem of distinguishing between 
oceanic-island and circumoceanic basalts, 
passes to a discussion of multigroup 
discriminants leading to a "cluster analy- 
sis" of lamprophyres with preliminary 
estimates of intra- and intercluster "dis- 
tances," and concludes with a brief 
examination of the relation between 
classification based on normative param- 
eters and classification based on discrimi- 
nants computed from chemical compo- 

The first section of part II presents new 
chemical and petrographic data obtained 
in the continuing study of Cenozoic 
peralkaline extrusives; the second section 
is essentially a correction of what is 
perhaps best summarized as a projection 
error, which appeared in an earlier dis- 
cussion of these rocks (Chayes and Zies, 
Year Book 61, fig. 38 and pp. 116-118). 
Further sections contain discussions of 
the relations between suites of Barth- 
Niggli and CIPW norms, of the distinc- 
tion between basalt and andesite, and of 
a possible heteromorphic relation between 
two varieties of lamprophyre. 

The report is written so that those who 
have little interest in rocks may ignore 
part II while those who are still unper- 
suaded of the advantages of statistical 
analysis may skip part I. Readers who 
have the courage and generosity to expose 
themselves to both parts will soon 
discover, however, that in this laboratory 
the distinction between statistical petrog- 
raphy and descriptive petrography is 
becoming increasingly difficult to main- 

I. Discriminant Functions and 
Petrographic Classification 

Oceanic- 1 stand and Circumoceanic Basalts 
F. Chayes and D. Mttais 

A comparison of the chemical compo- 
sition of Cenozoic volcanics "in and 
around the open oceans," in which the 
oceanic data were based entirely on 
materials collected from oceanic islands, 
was published during the report year (see 
abstract 1420 in Summary of Published 
Work, below). In earlier work the possi- 
bility that oceanic islands might provide 
a biased sample of oceanic lavas in general 
had been anticipated (Chayes, 1963, p. 
1531), and new data concerning the 
submarine basalts of the mid- Atlantic 
Ridge and the East Pacific Rise (Nicholls, 
1964; Engel and Engel, 1964a, b; Muir 
and Tilley, 1964) create a strong presump- 
tion that they do. Though hardly yet 
warranting blanket rejection of earlier 
speculations or the fabrication of new 
earth models, the 19 analyses of ridge 
basalts now available clearly bring to an 
abrupt conclusion the long era — dating, 
in fact, from the birth of petrography — 
during which generalizations about oce- 
anic volcanism were unavoidably based 
solely on data drawn from oceanic islands. 
Despite its rather general title, publica- 
tion 1420 is essentially a comparison of 
Cenozoic volcanics of the circumoceanic 
environment with those of the oceanic 

This work raised issues that could not 
be resolved by the methods initially 
applied. Accordingly, instead of moving 
from the oceanic-island and circumoce- 
anic Cenozoic volcanics to those of the 
shallow-sea and continental environ- 
ments, as originally planned, we have 
been attempting to resolve, by means of 
discriminant function analysis, two of the 
more important problems brought into 
focus by the earlier study. The following 
notes present a brief statement of these 
problems, a resume of the way in which 
discriminant function analysis may be 



brought to bear upon them, and a sum- 
mary of some results providing an answer 
to the first question. 

One of the more striking results to date 
is the remarkable difference of Ti0 2 
content in oceanic-island and circum- 
oceanic basalts. That oceanic-island ba- 
salts tend to be richer in Ti0 2 has been 
known for a very long time, but it does 
not seem to have been realized that the 
difference was either as large or as 
persistent as in fact it proves to be. On 
our data as originally assembled, in which 
any rock whose analysis listed less than 
54 per cent Si0 2 and yielded a Thornton- 
Tuttle index of less than 50 was classed 
as basalt, a discriminant set at Ti0 2 = 
1.75 assigns basaltic analyses to the 
appropriate geographic class with the 
rather extraordinary efficiency of 93.3 per 
cent. At present we are modifying our 
data file to permit compilation by rock 
names as listed in source references, and 
in a preliminary tabulation of 864 
"basalts as named in source references" 
the efficiency of a Ti0 2 discriminant at 
1.75 is again above 93 per cent. (More 
complex calculations with groups based 
on names used in source references have 
not yet been completed ; numerical results 
used below are based on the "machine 
definition" of basalt given above.) 

Whether a linear combination of some 
other oxide with Ti0 2 would provide a 
discriminant superior to Ti0 2 itself is 
obviously a matter of considerable inter- 
est, and it would also be useful to know 
whether some linear combination of 
oxides other than Ti0 2 might function as 
effectively as discriminants based on Ti0 2 . 
A method capable of answering the first 
of these questions would be capable, at 
least in principle, of answering the second. 
Such a method might also offer a new 
approach to study of the similar but more 
general questions involved in attempts to 
delineate petrographic provinces. 

From inspection of histograms and 
calculation of covariances it is obvious 
that no unweighted combination of oxides 
other than Ti0 2 will provide an effective 
discriminant, and it is a fairly safe guess 

that no unweighted combination of Ti0 2 
with other oxides will be materially more 
efficient than Ti0 2 alone. This is hardly 
more than a beginning, however, since it 
throws no light whatever on the possible 
effects of weighting, and it is precisely 
these effects that require examination 
before an answer to either question can 
be reached. Fortunately, information on 
the effects of weighting is provided by 
calculation of the appropriate discrimi- 
nant functions. If parent distributions are 
normal and parent variances homogene- 
ous, the coefficients of the discriminant 
function are in fact the weights that 
produce the most efficient classification 
based on a linear combination of the 
particular variables used in the calcu- 

An excellent introduction to the two- 
group discriminant function of Fisher 
(1936) is given by Hoel (1962). We 
present here a brief rationale of the pro- 
cedure, so that readers not concerned 
with the statistical theory or computa- 
tions may nevertheless appreciate their 
bearing on the petrographic argument. 

When every item in a sample can be 
classified, on the basis of properties 
ai, a2, • * • , ctu, as a member of either of 
two mutually exclusive classes, say &(a) 
and C2(a), it may be desirable to charac- 
terize the two groups by means of a 
second set of properties, the variables 
Xi, x 2 , - • - , x v . In our situation the a's 
are the properties which lead to the 
conclusion that an analysis was made on 
a basalt drawn from one of the two 
environments in question, and the x's are 
the observed amounts of the essential 
oxides recorded in the analysis. Because 
of interest in the relation — or lack of 
relation — between x's and a's we seek a 
function of the x's that maximizes the 
difference or "distance" between the 
groups originally established from the a's. 
For this purpose we form from each 
analysis a new variable, using subscripts 
to denote oxides, viz., 

z = XiflSi + X 2 x 2 + ■ • • + ^kX k 

where k < v, and solve for a set of X's that 



will maximize the ratio of between- to 
within-group variation of the z's, each 
type of variation being expressed as a 
sum of squared deviations. Given this set 
of X's, we next compute z for each analy- 
sis, either in the original set or, preferably, 
in data not used in the original compu- 
tation. For the two groups we then have 
the averages, z\ and z 2 , and if the parent 
ratio (21 — z 2 ) 2 /(Ni + N 2 )a z 2 is large 
enough — i.e., if the "distance" is great 
enough — a discriminant, 2, in the region 
Z\ < 2 < z 2 will efficiently reclaim the 
dichotomy based initially on the a's. If 
there is no evidence of heterogeneity of 
variance, the discriminant is estimated by 
the midpoint between the group averages, 
or 2 = (zi + 2 2 )/2, this value being chosen 
on the intuitively reasonable basis that 
the probability of misclassification is then 
the same for members of Ci(a) as for 
those of C 2 (a). If the variances are not 
homogeneous the same criterion is satis- 
fied by a discriminant placed at 
(<r 2 Zi + <t\Z 2 )/{<ji + <r 2 ), <tj denoting the 
standard deviation of z in group j. If the 
z's effectively reclaim the dichotomy 
based initially on the a's we conclude that 
the x's are significantly related to the 
as — or, in terms of our problem, that 
there is a significant association between 
geochemistry and geography. If, on the 
other hand, even the maximum "distance" 
is not large enough to make 2 an efficient 
discriminator, it is difficult to avoid the 
conclusion that the particular x's used are 
indifferent to the classification based on 

The classification established by a's 
and x's may be formed into a 2 X 2 
contingency table, as shown in table 14. 
If classification based upon a particular 
linear combination of x's perfectly 
duplicated the a classification the cell 
frequencies in the table would be F(ll) = 
N h F(12) = 0, F(21) = 0, F(22) = N 2 . 
The quantity 100(^11] + F[22])/(Ni + 
N 2 ), the percentage of "correct" classifi- 
cations made by the discriminant, is used 
here as a measure of the efficiency with 
which a classification based on variables 
Zi, x 2 , • - • , Xk proxies for the original 

TABLE 14. 2X2 Array, Showing Extent 
of Agreement of x Classification with 
a Classification 


C 2 (x) 








N 2 

(Ni + N s ) 

classification based on a h a 2 , • • • , a u . 

Returning now to the first of the two 
questions that prompted this study, if 
k = 1, 2 is simply a multiple of Xi, and the 
single coefficient, Xi, may be dropped. The 
discriminant then reduces to the midpoint 
between the group means for xi. If the 
variances are not homogeneous, this 
becomes (<ri 2 x n + cruXu) / (<ru + <t ]2 ), trtj 
denoting the standard deviation and Xij 
the mean of variable Xi in group j. A 
standard deviation correction proved 
essential in the discriminant based solely 
on Ti0 2 , and, since this oxide is always a 
major contributor to any more complex 
discriminant of which it is a member, the 
correction has been applied throughout. 
It is evident that (a 2 Zi + <jiZ 2 )/((ji + c 2 ) 
— > (21 + 2 2 )/2 as 1 0-1 — o- 2 1 — ■* 0, so that 
the correction influences the location of 2 
strongly only when the data suggest this 
is necessary. 

If the variables are uncorrelated or 
weakly correlated, those whose mean 
values differ most as between the groups 
will contribute most effectively to a 
discriminant. The largest group differ- 
ences are for MgO, Si0 2 , and A1 2 3 , and 
it is evidently among these that an 
effective increment to the discriminant 
based on Ti0 2 alone should first be 
sought. The discriminating efficiency of 
each combination of these three oxides 
with each other and with Ti0 2 is shown 
in table 15. It is evident from the table 
that combinations which include Ti0 2 are 
more efficient than those which do not 
and that the binary combination Ti0 2 - 
MgO is about as efficient as the quater- 
nary or any ternary including Ti0 2 . 
Sampling experiments have shown that 



TABLE 15. Efficiency of Discriminants 

Based on Various Combinations of Oxides of 

Ti, Mg, Al, and Si, as Indicated by Percentage 

of Correct Classification of the 356 Circum- 

oceanic and 579 Oceanic-Island Basalts from 

Which the Functions Were Computed 



Binary : 

























the superiority of the ternary combination 
Ti0 2 -MgO-Si0 2 over Ti0 2 -MgO is appar- 
ent rather than real, and the same is 
probably true also of the quaternary. 

Discrimination based on the combina- 
tion Ti0 2 -MgO, on the other hand, does 
appear to be somewhat more efficient 
than that based on Ti0 2 alone. Although 
an efficiency of 93.3 per cent can be 
obtained from the latter by inspection of 
histograms, that of the computed discrimi- 
nant is only 92.8 per cent. A difference of 
95.7 - 92.8 = 2.9 per cent, though not 
large in an absolute sense, represents a 
reduction of 2.9/7.2 = 40 per cent in the 
expected frequency of erroneous classifi- 
cation. This is an improvement worth 
having. In view of the weak correlation 
between Ti0 2 and variables other than 
those considered it seems safe to conclude 
that binary discriminants based on 
combinations of Ti0 2 with any of these 
variables will be less efficient. The answer 
to the first question thus appears to be 
that the most efficient binary discrimi- 
nant is based on Ti0 2 -MgO and that this 
combination is somewhat more efficient 

than Ti0 2 alone. The discriminant func- 
tion based on Ti0 2 and MgO may be 

z = Ti0 2 + 0.14 MgO - 2.80 

and sampling experiments indicate that 
if basalts from the two environments are 
classed as oceanic-island if z > and 
circumoceanic if z < no more than 1 
specimen in 20 will be misclassified. 

A definitive answer to the second 
question — viz., whether some combina- 
tion of oxides other than Ti0 2 might not 
discriminate as effectively as Ti0 2 — is not 
so readily obtained, if only because of the 
enormous amount of calculation that 
might be undertaken in connection with 
it. From the eight essential oxides other 
than Ti0 2 , no fewer than 2 8 - 1 = 255 
linear combinations can be formed. Ex- 
perience, common sense, and examination 
of sample statistics suggest that for most 
of these combinations discriminant func- 
tions are not worth calculating. For 
substantially the same sound reasons, 
however, the Ti0 2 distribution that forms 
the basis of this work remained unex- 
ploited for upward of half a century, so 
that it is perhaps wise to err on the side 
of caution in deciding which linear 
combinations to refrain from computing. 
In any event, work on this aspect of the 
problem is still fragmentary and will be 
reported in detail at a later time. 

Classification of Lamprophyres; a Possible 

Petrographic Application of Multigroup 

Discriminant Function Analysis 

D. Mttais and F. Chayes 

We summarized last year the first 
results of a new compilation of the avail- 
able chemical analyses of lamprophyres, 
tabulating average compositions for the 
six major varieties of these interesting 
dike rocks. For most of the essential 
oxides these average values are quite 
similar from group to group, the effect of 
an occasional abrupt shift in one oxide 
evidently usually being distributed over 
the remainder rather than compensated 
by a comparable shift of opposite sign in 



some other oxide. Variation within each 
type is extremely large, furthermore, and 
the lamprophyre that can be unequivo- 
cally identified from its raw analysis is 
rather unusual. 

In the lamprophyres the original 
grouping — the analogue of the a classifi- 
cation of the preceding section — is essen- 
tially mineralogical. We report here the 
first stages of a study one of the objectives 
of which is to determine the extent to 
which multigroup discriminant functions 
based on chemical composition can be 
made to proxy for this basically mineral- 
ogical classification. 

Multigroup discriminant procedures 
are described in great detail by Rao 
(1952), and excellent accounts of possible 
applications in other sciences are to be 
found in Rao and Slater (1949) and in 
Reyment (1963). For readers who, like us, 
have little previous experience with 
modern discriminatory techniques, the 
major practical distinction between two- 
and multigroup discriminants is just that 
the former can be conveniently defined, 
computed, and interpreted without ma- 
trix algebra and the latter cannot. For 
both types, however, the underlying 
objective is the same, viz., to maximize 
the ratio of between- to within-group 
dispersion — the so-called "distance" — 
between any pair of groups. 

Denoting by co;y the element of the 
total within-group dispersion matrix for 
any pair of variables i, j, we wish for each 
pair of groups a set of X's such that 

Z> 2 = £ XX*; 

AiO>n + X2CO12 + 
X1CO21 + X2CO22 + 

AlO>j,l + X2CO p 2 + 

• • ApCOip — Ui 

• ' ApC02p == di 

XpWpp — d 1 

where, as in the two-group discriminant, 
\ q is a weighting coefficient specific to 
some one of the p variables and d q is the 
difference between the average values of 
variable q in the two groups. The 
Mahalanobis "generalized distance" be- 
tween any two groups is then defined as 


= Z T, 0,^' did 3 

i=i y=i 

where D 2 , the X's, and the d's are specific 
to the two groups in question and co ij is 
the ijth element of the inverse of the 
within-group dispersion matrix of the 
entire array. The term to the right of the 
second equality sign can be computed 
directly from the data, so that an estimate 
of D 2 does not require prior computation 
of the actual discriminant function. In 
view of its definition, however, it is evi- 
dent that if D 2 is large a discriminant 
based on the same set of variables will 
effectively proxy for the as of the 
original classification, whereas if it is 
small such a discriminant will be ineffi- 
cient. The computation of D 2 for each 
pair of groups is thus a useful first step. 

For the six large groups of last year's 
report plus two new ones — the non- 
feldspathic ouachitites and alnoites 11 — 
arrays of D 2 values have been computed 
for several sets of variables. D 2 values are 
invariably small if based on variables 
other than Si0 2 , MgO, and K 2 0, whereas 
little further enlargement of D 2 over the 
values yielded by this set results if other 
variables are added to it. The Mahala- 
nobis "generalized distances" obtained 
from the linear combination Si0 2 -MgO- 
K 2 are shown in table 16. In view of 
doubts about the randomness of the 
sampling and ignorance about the nor- 
mality of the parent distributions of the 
variables involved, we refrain from 
numerical tests for significance. 

It is evident that certain of the inter- 
group distances are small and others 
large. Forming "clusters," in the fashion 
suggested by Rao, of groups separated 
from each other by short distances, and 
averaging each set of "within"- and 
"between"-cluster distances, leads to the 
result shown in table 17, in which the 

11 For definitions of the various types of 
lamprophyre see, for instance, Troger (1935). 



cluster "diameters" are shown in italics 
along the leading diagonal and the other 
entries are average distances between 
clusters denoted by row and column 
headings. (Since cluster I contains only 
one group its diameter is undefined.) 

Inclusion of other variables slightly 
enlarges cluster diameters but has much 
less influence on intercluster differences; 
for all combinations so far tested its effect 
is negligible. Moving in the other direc- 
tion, it is interesting to note that deletion 
of one of the variables, K 2 0, also seems 
to make rather less difference than might 
be expected. For the linear combination 
(SiCVMgO) the diameter of cluster II is 
reduced and the distance between clusters 
I and II becomes small in relation to the 
reduced diameter. For this combination 
minettes would accordingly be added to 
cluster II, and cluster I would vanish, but 
there are no other changes of importance. 

There is thus a strong suggestion that 

for distinguishing between closely related 
varieties mineralogy is more useful than 
chemical composition. Kersantites and 
vogesites almost certainly cannot be 
efficiently distinguished by means of 
discriminant functions based on chemical 
composition, for instance, but if one is 
told the names of the minerals in a 
specimen one can decide at once, without 
benefit of discriminant functions, whether 
it is a kersantite or a vogesite. In view of 
the fact that the basic definitions are 
mineralogical this is hardly a surprising 
result. (The relation between vogesite and 
kersantite is discussed in more detail in a 
later section of this report.) 

Normative color index is low in cluster 
II, intermediate in cluster III, and high 
in cluster IV. The average compositions 
of these clusters differ considerably, as 
may be inferred from the group averages 
shown in Year Book 62 (p. 157, table 11). 
The range of composition within each 

TABLE 16. Values of Mahalanobis D 2 for the Linear Combination (Si0 2 -MgO-K 2 0) 

No. of 

Analyses : 


Spessart- Campton- Monchi- Ouachi- 

tite Minette V °S esite ite ite quite tite 

103 66 30 45 78 61 10 


























TABLE 17. Intra- and Intercluster Average D 2 for the Linear Combination (Si0 2 -MgO-K 2 0) 



































group is so large, however, that with the 
possible exception of Si0 2 no oxide or 
unweighted combination of oxides will 
efficiently classify individual specimens. 
From the intercluster distances shown in 
table 17 it nevertheless seems possible 
that a properly chosen set of discriminant 
functions would perform with suitable 
efficiency. We are now attempting to 
choose such a set and to devise a test of 
efficiency in the multigroup situation 
comparable to that described, in the 
preceding section, for a simple dichotomy. 

Discriminant Function Coefficients 

and Normative Calculations 

F. Chayes 

Although a completely general state- 
ment of the relation between normative 
calculations and the coefficients of the 
discriminant function calculated from the 
relevant oxides is remarkably complex, in 
specific situations this relation may be 
both simple and very informative. 

In oversaturated meta- or peralumi- 
nous rocks, for instance, the amount of 
normative alkali feldspar is simply 

or + ab = 5.91 K 2 + 8.46 Na 2 

A similar abbreviated calculation of an 
requires a term in A1 2 3 for metaalumi- 
nous and in CaO for peraluminous 
analyses. Specifically, if, in molar amounts, 

(K 2 + Na 2 0) < A1 2 3 < 

(K 2 + Na 2 + CaO) 

the total normative feldspar is 

or + ab + an 
= 2.96 K 2 + 3.97 Na 2 + 2.73 A1 2 3 

whereas if 

(K 2 + Na 2 + CaO) < A1 2 3 

or + ab + an 
= 5.91 K 2 + 8.46 Na 2 + 4.96 CaO 

(The distribution of CaO to accessories 
has been ignored.) It is to be noted that 
all three equations are linear combina- 
tions, and it follows at once that if a 
discriminant function based on the rele- 
vant combination of oxides, (K 2 0, Na 2 0, 

CaO, A1 2 3 ), fails to perform with suit- 
able efficiency there is no need to test 
specific hypotheses that any of the 13 
distinguishable sets based on sums or 
differences of one or more of the norma- 
tive feldspars will do any better. None of 
them will be superior to the discriminant 
calculated directly from the oxides, for 
the coefficients of this discriminant pro- 
vide the maximum efficiency obtainable 
under the circumstances. 

Whenever a normative parameter can 
be expressed as a linear combination of a 
particular set of oxides, taxonomic hy- 
potheses based on this parameter are 
among the family of such hypotheses 
tested by the discriminant based on the 
oxides. In principle, I believe, any 
normative parameter may be expressed as 
a linear combination of oxides, but the 
calculations are sometimes exceedingly 
prolix. They would often be much simpli- 
fied by transformation to molar per- 
centages; in this event the discriminant 
calculations would also be carried through 
in the same form. For anhydrous or uni- 
formly hydrous rocks this introduces no 
difficulty. For suites markedly variable in 
water content, however, transformation 
to molar percentages assures either that 
an enormous importance is attributed to 
H 2 or that from the outset one works 
with a series of projections rather than 
with original observations. 

It is of course possible to calculate a 
discriminant from the normative param- 
eters themselves, and with electronic 
computation it actually makes little 
practical difference whether the discrimi- 
nant is calculated from oxides or norms. 
Calculated from the oxides, each discrim- 
inant tests a whole class of taxonomic 
hypotheses, including many — in particu- 
lar those not susceptible of ready norma- 
tive or mineralogical rationalization — 
that have not been specifically recognized 
or articulated. Considering the history 
and present condition of systematic 
descriptive petrography, this is probably 
a considerable advantage. 

To the questions one attempts to 



answer by means of discriminant func- 
tions — as to most scientific questions — 
there is only one completely unambiguous 
answer, viz., "no." A discriminant func- 
tion that performs inefficiently auto- 
matically throws out of court all other 
linear combinations of the same variables. 
But one that performs efficiently may be 
rather difficult to interpret. For example, 
an efficient discriminant based on a set of 
N variables is not to be construed as 
evidence of the efficiency of a discriminant 
based on any particular subset containing 
less than N of these variables. On the data 
described in the first section, for instance, 
a discriminant based on the combination 
(CaO, Na 2 0, K 2 0, A1 2 3 , Si0 2 , Ti0 2 ) 
proved 95.3 per cent efficient, but one 
based on the first five of these oxides is 
only 60.1 per cent efficient, and the 
efficiency of one based on the first three 
is only 52.7 per cent. These numbers are 
to be compared with a "zero" at 50 per 
cent, which is what would be expected 
from a scientist who made his decisions 
by flipping coins instead of calculating 
discriminant functions. Obviously most 
of the work of the 6-variable discriminant 
is done by Ti0 2 despite the fact that the 
discrimination is between what are tra- 
ditionally regarded as alkaline and calcal- 
kaline basalts. As with so many tech- 
niques made practical by the ready 
accessibility of high-speed computation, 
discriminant function analysis is a supple- 
ment to but not a substitute for contem- 

II. Chemical and Optical 

Notes on Some Mediterranean Comendite 

and Pantellerite Specimens 

F. Chayes and E. G. Zies 

As part of a continuing study of the 
petrography of Cenozoic acid volcanics, 
work was completed during the report 
year on three previously undescribed 

Hyalopantellerite from Cuddia Gadir, 
Pantelleria (specimen 44B5, collected by 

F. R. Boyd). This is a clear black obsidian 
with about 12 per cent of phenocrysts. 
The phenocrysts are virtually the only 
perceptibly crystalline material in the 
rock, the groundmass being almost per- 
fectly glassy. A modal analysis is shown 
in table 18, chemical analyses of the bulk 
sample and of feldspar phenocrysts 
separated from it are shown in table 19, 
and normative data are given in table 20. 
For a number of reasons this specimen 
is chiefly of interest in comparison with 
Washington's type hyalopantellerite, from 
Gelkhamar, for which comparable data 
were recorded in Year Book 61 (pp. 112— 
116, specimen PRC 2000). In the new 
specimen quartz phenocrysts are con- 
siderably more and feldspar phenocrysts 
somewhat less abundant, but the rounded, 
corroded appearance of the quartz is 
much the same in both specimens, and so 
too are the habit, optical properties, and 
chemical composition of the sanidine. 
Glass is by far the most abundant con- 
stituent of both specimens; in PRC 2000 
it is heavily charged with identifiable 
microlites of feldspar, cossyrite, and 
acmite, but in 44B5 it is virtually free of 
recognizably crystalline matter. In chem- 
ical composition and norm the two rocks 
are very similar; both are rich in CI, and 
in both there is a considerable molar 
excess of alkalies over trivalent metals. 
The locality of the new specimen is near 
the northeastern edge of the island, 

TABLE 18. Modal Analyses 

(For identification of specimens see text or 

notes to table 19.) 



















Brown hornblende 
















TABLE 19. Chemical Analyses 








Si0 2 








Ti0 2 







A1 2 3 








Fe 2 O s 






























Na 2 








K 2 
















Zr0 2 




P 2 5 





v 2 o 5 







H 2 0~ 
H 2 0+ 





} None 

} 0.03 



S0 4 








} 0.03 








= Cl 







44B5. Hy alopan tellerite, Cuddia Gadir, Pantelleria. 

44B5F. Feldspar from 44B5. 

ZPTl. Trachyte, Cantina la Croce, Pantelleria. 

ZPT1P. Pyroxene from ZPTl. 

ZPT1M. Magnetite from ZPTl. 

39B3. Comendite, Capo Sandolo area, San Pietro. 

39B3F. Feldspar from 39B3. 

Mineral separations and analyses by E. G. Zies. 

* Sought but not found, 
f Probably present. 
t Soluble in water. 
§ Insoluble in water. 

several kilometers east of Gelkhamar. 
The geography of the intervening terrain 
is such that the two occurrences are 
almost certainly independent at the 
present erosion level. 

Trachyte from near Cantina la Croce, 
Montagna Grande, Pantelleria (specimen 
ZPTl, collected by Zies and G. Vianelli 
of the University of Palermo). Unlike the 
hyalopantellerites, which are notable for 
their apparent petrographic simplicity, 
this is an astonishingly complex rock. It 
consists of a purplish matrix, very fine 
grained and partly glassy, in which are 

set abundant irregularly shaped inclu- 
sions and numerous gray to greenish 
yellow feldspar tablets. The vitreous 
luster of the phenocrysts contrasts sharply 
with the dull cast of the matrix. The 
inclusions are readily detectable in hand 
specimen. Their abundance and extreme 
variation in size, shape, and modal 
composition are among the most striking 
features of the rock in thin section. A 
modal analysis based on three large thin 
sections is shown in table 18, but the 
reader is cautioned that the distribution 
of inclusions is erratic and the distinction 



TABLE 20. Norms of Rock Analyses 

(For identification of specimens see text or 

notes to table 19.) 

















































between coarse-grained inclusions and 
glomeroporphyritic aggregates of pheno- 
crysts is sometimes rather arbitrary. 
Chemical analyses of the rock and of 
pyroxene and highly titaniferous "mag- 
netite" separated from it are shown in 
table 19. Table 20 includes a norm 
calculated from the rock analysis. 

Except for plagioclase and a lavender- 
colored pyroxene, possibly titanaugite, 
which are confined to certain of the 
inclusions, the same minerals — alkali 
feldspar, pale green diopsidic augite, 
magnetite, and, rarely, olivine — occur 
throughout. Although Q appears in the 
norm, no modal quartz has been found. 

The alkali feldspar is a mixture of 
sanidine and anorthoclase, the first term 
denoting an untwinned or simply twinned 
phase and the second a phase in which 
polysynthetic twinning is always and an 
imperfect microcline-like grid is some- 
times present. The sanidine of this rock 
can always be made to take a stain, 
though the result sometimes leaves much 
to be desired in the way of smoothness 
and continuity. Anorthoclase usually 
stains as deeply as sanidine. In two of the 
thin sections most of the clearly identi- 
fiable euhedral phenocrysts are sanidine. 
In the third section phenocrysts of 

(stained) polysynthetically twinned anor- 
thoclase are abundant. 

Complexly twinned plagioclase is al- 
most entirely confined to inclusions or 
large glomeroporphyritic aggregates; it is 
nearly always mantled, usually by sani- 
dine, less commonly by anorthoclase. The 
demarcation between stained mantle and 
unstained core is sharp. As has already 
been noted, the distinction between 
coarse-grained inclusions and clusters of 
phenocrysts is far from clear; quite 
possibly many now isolated euhedral 
tablets of feldspar may have been derived 
initially from fragmented inclusions. 
Plagioclase-bearing inclusions are scarce 
in two of our sections and are not abun- 
dant in the third. In this respect the 
thin-section sampling is probably inade- 
quate; attempts to obtain a pure alkali 
feldspar concentrate by magnetic and 
heavy-liquid separation were finally aban- 
doned because locked particles of plagio- 
clase proved impossible to remove and too 
abundant to ignore. 

Except for lavender augite, which 
occurs only in a few inclusions, the 
pyroxene of the rock is a pale green, non- 
pleochroic, monoclinic variety which 
looks very like diopside. It contains rather 
more soda and alumina than can be 
accounted for by contamination with 
feldspar, and a considerable molar excess 
of (MgO + FeO) over CaO. This pyrox- 
ene occurs in abundance as very fine 
grains in the groundmass, commonly as 
separate phenocrysts considerably smaller 
than those of feldspar, occasionally as 
contaminants of feldspar, and in crystals 
of all sizes in inclusions. It is by all odds 
the principal dark silicate and in fact the 
only one of any quantitative importance. 
Olivine occurs sparingly in inclusions, in 
the groundmass, and occasionally as 
small phenocrysts. It is considerably less 
abundant than a highly magnetic opaque 
mineral, which is also found both as 
phenocrysts and in the groundmass, and 
for whose examination we are indebted to 
D. H. Lindsley, P. Ramdohr, and C. W. 



Ramdohr finds that this material is 
essentially a mixture of magnetite and 
maghemite with a little ilmenite and very 
rare grains of sulfide, lepidocrocite, and, 
possibly, chrome-spinel. Lindsley reports 
that X-ray powder spectra show only 
peaks corresponding to the spinel struc- 
ture; these are sharply resolved but 
slightly skewed toward the high-angle 
side, the skew apparently being the only 
evidence in the pattern of the maghemite 
observed optically. In table 21 interplanar 
spacings observed by Lindsley are com- 
pared with those yielded by a least- 
squares refinement carried out by Burn- 
ham, from which the cell edge is estimated 
as 8.479 ± 0.002 A and the volume as 


609.6 ± 0.5 A 3 . Lindsley points out that 
according to the work of Akimoto, 
Katsura, and Yoshida (1957) these cell 
dimensions are consistent with the bulk 
composition (table 19) recalculated to 
100 per cent (FeO + Fe 2 3 + Ti0 2 ). 

The mineralogical data already re- 
viewed suggest that most of the inclusions 
in the Cantina la Croce specimen are 
members of a suite of hypabyssal rocks 
closely related to the trachyte. Most of 
them, and particularly the highly feld- 
spathic ones, are rich in apatite. This 
mineral is of markedly acicular habit, and 
similar needles are not infrequently found 
in the host rock. 

Despite its high alkali content and its 
close association with extraordinarily 

TABLE 21. Observed and Calculated 

Interplanar Spacings of Magnetic Opaque 

Mineral Separated from Specimen ZPT1 

(Separation by Zies, d obs by Lindsley, 

dcaic by Burnham) 



















1511 j 






peralkaline rocks, the trachyte of Cantina 
la Croce is not alkaline in the sense of 
Holmes. The rather high Fe 2 3 /FeO ratio 
suggests extensive oxidation, but the 
subalkaline character persists even if all 
the Fe 2 3 is converted to FeO before 
calculation of the norm. From Washing- 
ton's account and map, most of the 
trachyte of Pantelleria is of this same 
character. For his principal trachyte map 
unit, the so-called "Gibele type" (Wash- 
ington, 1914), he presents three norms; 
all three, like the Cantina la Croce norm, 
contain Q, hy, and an, and accordingly 
lack ne, ac, and ns. 

Comendite from hill 1^2, near Capo 
Sandolo, San Pietro, Sardinia (specimen 
39B3, collected by Chayes). A blue-gray 
streaky rock with strong flow texture and 
numerous phenocrysts of bipyramidal 
quartz and tabular sanidine, this speci- 
men is of immediate interest primarily in 
comparison with a companion specimen 
(39B2), described in Year Book 61. The 
collecting sites are separated by only a 
few hundred meters, and the specimens 
are very similar. Under the microscope 
the principal distinction is the degree of 
devitrification; the groundmass of 39B2 
is almost completely spherulitic whereas 
that of 39B3 is nearly everywhere iso- 
tropic and shows little suggestion of 
devitrification. Despite this rather strik- 
ing contrast the rocks are rather similar 
in composition; though on the whole 
rather small, however, the differences 
appear to be compatible with the notion 
that the spherulitic material is essentially 
a hydro thermal modification of the glass. 
Specifically, total H 2 is 1.13 per cent in 
the devitrified rock as compared with 0.59 
in the relatively unaltered one, the 
devitrified rock is 2 per cent richer in 
Si0 2 , its Fe 2 3 /FeO ratio is almost 50 per 
cent greater, and its total Fe content 
almost 50 per cent less. The norm of the 
glassy rock contains 0.27 per cent ns and 
10.33 per cent ac; that of the devitrified 
specimen lacks ns and shows only 1.8 
per cent ac. 

The normative feldspar of each rock is 



about 5 per cent richer in or than the 
phenocrysts are, but the conventions of 
the normative calculation preclude any 
simple interpretation of this relation; the 
Na 2 0/K 2 ratio is considerably less in 
the devitrified rock, for instance, but its 
normative feldspar is richer in ab. Chem- 
ical analyses of 39B3 and of the feldspar 
extracted from it are shown in table 19, 
and the norm of the rock analysis in table 
20. Similar data on the devitrified speci- 
men (39B2) are given in Year Book 61 
(pp. 112-116, tables 6, 7, and 8). 

Descriptive and Genetic Significance 

of Normative ns 

F. Chayes 

Projections in which CIPW normative 
molecules are used as components have 
stimulated much fruitful descriptive and 
experimental work, and inferences based 
upon them have proved correct often 
enough so that users of this particular 
system of petrographic calculation some- 
times lose sight of the distinction between 
the norm as a descriptive device and its 
application as a genetic indicator. The 
second is what we wish it were and what 
it has sometimes proved to be; the first 
is what it is and was intended to be. 

An example of overextrapolation from 
a diagram based on normative compo- 
nents occurred in Year Book 61 (Chayes 
and Zies, pp. 112-118, fig. 38 and 
accompanying discussion), in which 
Q:or:ab ratios of volcanics from Pan- 
telleria and Isola San Pietro, plotted in 
the central region of a ternary diagram, 
were compared with the normative 
or /(or + ab) ratios of phenocrysts from 
the same rocks, plotted along the or-ab 
edge. In the comendites of San Pietro the 
difference between or content of norma- 
tive alkali feldspar in phenocrysts and 
whole rock was negligible ; in the peralka- 
line obsidian of Pantelleria, on the other 
hand, it was much higher in the projected 
value of the rock as a whole than in the 
phenocrysts. From this it was suggested 
that "fractionation of the Pantellerian 

type would involve extensive end-stage 
enrichment in potassium, whereas frac- 
tionation of the comenditic type would 
proceed with no notable shift in the Na/K 

Carmichael (1962) had already noted 
that the K/Na ratio in feldspar pheno- 
crysts varied little and was considerably 
lower than in projected bulk compositions 
for three other Pantellerian specimens, 
and he and MacKenzie (Carmichael and 
MacKenzie, 1963) subsequently used the 
same projection scheme to rationalize 
experimental results concerning crystal- 
lization paths in charges whose compo- 
sitions, except for H 2 0, could be repre- 
sented by components having the formu- 
las assigned the CIPW normative mole- 
cules or, ab, Q, ns, and ac. Bailey and 
Schairer (1964a) now draw attention to 
the curious circumstance that, although 
in the part of the bulk composition pro- 
jected into the system Q-or-ab the K/Na 
ratio is indeed considerably higher than 
in the feldspar phenocrysts, in the 
pantellerite analyses cited by both sets of 
authors it is actually a little lower. The 
fractionation trend proposed by Chayes 
and Zies is thus impossible unless the 
entire system changes composition — as, 
for instance, by loss of volatiles during 
crystallization — and this was no part of 
the original proposal. 

It is also no part of the rationalizations 
upon which the CIPW norm calculations 
are based, although the inconsistent 
materials balance noted by Bailey and 
Schairer is in fact a direct consequence of 
careless interpretation of the CIPW con- 
ventions regarding the partition of alka- 
lies between feldspar and metasilicate 
molecules. It is tedious but not difficult 
to show that if there is no differential 
partition of alkalies between these mole- 
cules the inconsistency disappears ; but so, 
alas, does the proposed pantellerite frac- 
tionation trend. 

In an analysis in which (Na 2 + K 2 0) 
> A1 2 3 , the CIPW convention is to 
combine Na 2 with Fe 2 3 to form ac, and 
if (A1 2 3 + Fe 2 3 ) > (Na 2 + K 2 0), as 



nearly always happens, this ends the 
matter. If, however, (A1 2 3 + Fe 2 3 ) 
< (Na 2 + K 2 0), remaining Na 2 is 
combined with Si0 2 in ns. At each stage 
the test is of available total alkali but the 
action taken as a result of the test affects 
only the soda. The decision to allot only 
Na 2 to ac is based on extensive and 
persuasive mineralogical evidence. But 
what is the mineralogical (or other) 
justification for allotting no K 2 at all to 
the metasilicate molecule? 

The ns molecule, regarded as femic by 
the authors of the CIPW system, is a 
purely formal artifice devised to ration- 
alize complications introduced by the 
occasional occurrence of highly peralka- 
line femic minerals such as arfvedsonite 
(see Cross et al., 1902, p. 644). The 
peralkaline femic minerals are often rich 
in Ti0 2 , however, and titaniferous mafic 
silicates other than acmite, cossyrite, and 
augite are often relatively rich in K 2 0. 
Potassium metasilicate (ks) is abundant, 
for instance, in the norms of the so-called 
lamproites, which contain very little 
Na 2 0, and it is perhaps noteworthy that 
in certain of these rocks a mineral 
originally called rutile has now been 
identified as potassium titanate (Prider, 
1960). Evidence for the K-Ti association 
is neither abundant nor conclusive, and 
the potential complication of normative 
calculation by the introduction of a 
potassium titanate molecule seems rather 
forbidding. In the pantellerites under 
discussion, however, normative ns is 
considerably greater than could be 
accommodated in any known peralkaline 
dark silicate, so that the mineralogical 
guideposts so ingeniously utilized in the 
standard CIPW calculation are of no 
avail. Experimental evidence, reviewed 
by Tuttle and Bowen (1958, pp. 85-87), 
indicates that in fractional crystallization 
the tendency toward enrichment of end- 
stage liquids in alkali silicates is as strong 
in the systems K 2 0-Al 2 3 -Si0 2 ± H 2 as 
in the systems Na 2 0-Al 2 3 -Si0 2 ± H 2 0. 
In rocks of this kind there thus seems to 
be as much — or as little — reason to com- 

pute normative metasilicate out of K 2 
as out of Na 2 0. 

To do so in the present instance, as in 
all analyses in which molar A1 2 3 > K 2 0, 
requires abrogation of the simple and 
basic CIPW rule that assigns all K 2 to 
A1 2 3 . When the norm is used as a 
cataloguing device, ad hoc modifications 
of this sort should certainly be avoided. 
When normative computations are used 
as a guide to petrological interpretation 
leading to unrealistic results, however, it 
is desirable to examine alternative com- 
bining rules. 

For the conversion of most rock 
analyses to CIPW norms the metasilicate 
and chloride molecules are not used at all, 
and in the system as a whole they are 
little more than nuisance parameters. 
When the CI content and the molar excess 
of alkalies over available R 2 3 are trifling, 
as they usually are, no set of combining 
conventions concerning these molecules 
can seriously influence the values of other 
normative molecules. 

With a CI content of more than 0.7 per 
cent and an excess of alkalies over R 2 3 
sufficient to generate more than 5 per cent 
ns in the norm, however, the conventions 
governing assignment of alkalies to hi and 
ns will powerfully affect the values of or 
and ab and will even exert appreciable 
influence on Q. Table 19 and figure 71 
illustrate this effect on analyses of the 
type hyalopantellerite of Gelkhamar 
(Zies, 1960, PRC 2000) and a new speci- 
men collected by F. R. Boyd from Cuddia 
Gadir (specimen 44B5, described in the 
preceding section of this report). The 
solid triangles along the base show the 
or I {or + ab) ratio of feldspar phenocrysts 
in these two rocks. The open circles show 
the ratio Q:or:ab for bulk compositions, 
as obtained from the CIPW combining 
conventions. The solid circles show the 
same ratio computed by assigning Na 2 
to ac but then distributing the remaining 
alkalies, without further differential par- 
tition, among feldspar, metasilicate, and 
chloride molecules. The contrast between 
the two projections is striking. In this 




Fig. 71. CIPW (open circles) and modified norm (solid circles) projections of PRC 2000 and 
44B5 in Q-or-ab ternary diagram. Compositions of feldspar phenocrysts (solid triangles) shown 
along or-ab edge. fe 

particular rock the metasilicate and 
chloride molecules are wholly and the 
feldspar molecules largely occult, so there 
is no possibility of deciding which pro- 
jection is preferable on the basis of a 
norm-mode comparison. 

In the modified projection the norma- 
tive feldspar of the bulk composition is 
slightly more albitic than that of the 
phenocrysts, as might be expected, other 
things being equal, from the circumstance 
that the ratio K 2 0/Na20 is somewhat 
lower in the rocks than in the pheno- 
crysts. In the standard CIPW projection, 
on the other hand, the normative feldspar 
of the bulk composition is much richer in 

or than the normative feldspar of the 
phenocrysts, and the resulting slope of the 
tie lines prompted the suggestion of a 
fractionation trend leading to end-stage 
enrichment in K 2 0. It should by now be 
clear that this apparent relation between 
rock composition and composition of 
phenocrysts is completely implicit in the 
CIPW combining conventions and is not 
in any sense evidence for the proposed 
fractionation trend. Support for this 
hypothesis must be sought in other evi- 
dence, and one may wonder what kind of 
evidence can make it seem reasonable 
that the extraction of crystals having a 
K 2 0/Na 2 ratio equal to or larger than 



that of a parent magma can enrich the 
magmatic residue in K 2 relative to 
Na 2 0. 

On the Relation between Suites of CIPW 

and Barth-Niggli Norms 

F. Chayes and D. Mttais 

The best known normative calculating 
procedures are those of the CIPW system, 
widely used by petrographers in this 
country, and the Niggli katanorm, now 
rather generally used in continental 
Europe. (The "Lacroix norms" of French 
petrology are CIPW norms, though in 
France, as in the United States, the 
classification based upon them never 
became popular. After abjuring all forms 
of normative statement for many years, 
the British, largely under the influence of 
Tilley, have gradually opted for CIPW. 
Among Japanese petrologists, on the 
other hand, one system seems about as 
popular as the other, and few petro- 
graphic papers lack norms.) 

As originally proposed, the two systems 
differ materially in combining conven- 
tions, and whereas the Niggli katanorm 
is cast in units of cation per cent on an 
anhydrous basis, the CIPW norm is in 
weight per cent with no correction for 
water content. In Year Book 62, to which 
the interested reader is referred for 
details, we reported good agreement be- 
tween parameters in whose estimation the 
two systems employed common combin- 
ing conventions but poor agreement or 
violent disagreement when a normative 
parameter assigned the same name was 
actually estimated by different rules in 
the two systems. In highly silicic analyses, 
for instance, the rules for calculating or 
are the same in both systems, and good 
agreement is to be expected; when silica 
is so low that the CIPW norm contains 
Ic, however, the rules governing the 
formation of or are no longer the same in 
the two systems (Ic being undefined in the 
Niggli system), and orcipw is so different 
from or N iGGLi that the use of a common 
symbol for the two is misleading and very 
nearly specious. 

In general, the more nearly identical 
the combining rules the better the agree- 
ment to be expected. The Niggli system 
is designed to permit calculation of norms 
appropriate to a variety of P-T facies, 
and there is in fact no firmly fixed 
standard norm for any particular facies. 
Some years ago Barth proposed a norm 
calculated in cation per cent, as in the 
Niggli system, but with the combining 
conventions of the CIPW system. In the 
purposely loose structure of the Niggli 
system this is a perfectly legitimate 
variant and probably should be referred 
to as the "Barth-Niggli katanorm." In 
practice such norms are usually called 
simply "molecular"; the reader accus- 
tomed only to the CIPW calculating 
technique may like to know that, if each 
"molecular amount" is divided by the 
sum of all of them, the normative amounts 
formed by the usual additive combina- 
tions of these molecular percentages (but 
without the final reconversion to weight 
percentages) will be a Barth-Niggli 
"cation" or "molecular" norm. We pre- 
sent here a brief summary of results 
obtained in a rather extended comparison 
of "molecular" norms with the standard 
CIPW or "weight" norms. 

In the first place, it is to be noted that 
there will be complete qualitative agree- 
ment between weight and molecular 
norms. Accordingly, decisions about no- 
menclature or taxonomy based on the 
presence or absence of particular norma- 
tive molecules will be the same whichever 
system is used. Further, although the 
actual amounts of normative molecules 
differ, in every suite of analyses tested 
the correlation between individual param- 
eters in weight and molecular norms is 
very high; the lowest value of p 2 so far 
recorded is 0.992. 

It is thus evident that linear conversion 
from one norm to the other may be per- 
formed with high precision, though the 
optimum conversion equations differ 
somewhat from suite to suite. Letting 
Y = molecular norm mt and X = weight 
norm mt, for instance, and writing the 



conversion equation as Y = aX + b, we 
find that in 11 suites of analyses 0.73 < a 
< 0.78 and -0.12 < b < 0.01; about 
this same range of slope and intercept 
holds also for il. 12 

Of the salic minerals, an, Ic, and or are 
usually in quite good agreement, but Q is 
appreciably lower and ne and ab are 
appreciably higher in molecular norms 
than in corresponding weight norms. Q, 
ne, and ab differences are usually of the 
order of 1 to 3 per cent, enough to be 
bothersome but rarely enough to affect 
materially inferences based on ratios of 
petrological interest calculated from these 
parameters. For Q, or, and ab in a typical 
Adirondack granite, for instance, the 
weight norm gives 28.62, 21.75, 36.42, and 
the molecular norm 26.77, 21.95, 39.07. 
From the weight norm the ternary ratio 
is Q = 32.9, or = 25.1, and from the 
molecular norm it is Q = 30.5, or = 25.1. 

For dark silicates the relation between 
weight and molecular norms is much less 
consistent. Depending on the Mg/Fe 
ratio the amounts of hy, di, and ol in the 
molecular norm may be either more or 
less than in the weight norm; di values 
differing by more than 1.5 are rare, but 
ol values differing by more than 2.5 are 
common. On the whole, however, it seems 
safe to say that speculations based on the 
amounts of standard norm minerals, or on 
ratios formed from these amounts, will be 
little affected by whether the norms 
involved are weight or molecular. 

If we are concerned with the composi- 
tion rather than the amount of a norma- 
tive parameter, however, the situation 
may be quite different. The important 
femic normative molecules are treated as 
solid solutions, and one has the option of 

12 It is clear that the conversion relation cannot 
be linear over the full composition range. For any 
normative molecule the weight and molecular 
norms will be in perfect agreement at and 100 
per cent; the discrepance between the two 
increases continuously to a maximum at some 
intermediate but not necessarily central value. 
In most suites of related analyses, however, the 
observed range is small enough so that the 
curvature of the conversion relation is negligible. 

considering the feldspars in the same way, 
though the CIPW system provides no 
conventions governing the partition of ab 
between alkali feldspar and plagioclase. 
It is quite possible that, although molec- 
ular- and weight-norm estimates of the 
amount of such a constituent are in 
reasonable agreement, they differ signifi- 
cantly about its composition. Normative 
olivine in the lavas of Vesuvius provides 
an excellent example. In 21 olivine- 
normative analyses ol ranges from 1.27 to 
9.64 in the weight norms and the average 
difference between weight and molecular 
norm values is only 0.25, the molecular 
norms being invariably higher. Fo content 
of ol in the weight norms ranges from 55 
to 98 per cent with only three values 
greater than 77 per cent, but on the 
average the fo content of ol is 6.4 per cent 
higher in molecular norms. In the olivine 
normative lavas of Velay, France, on the 
other hand, the average difference in fo 
content of ol is about the same (5.5 per 
cent) but molecular-norm ol averages 1.41 
more than weight-norm ol. The relation 
between fo content of ol in molecular and 
weight norms is of course functional, viz., 

/Oweight = /O mo le/ (1.448 — 0.448 jfomole), 

and quite independent of other compo- 
sitional characteristics of the analysis. 
The amount of ol in the two types of 
norms, however, is not independent of the 
rest of the analysis. In any event, 
consistency is quite evidently not one of 
the striking characteristics of comparisons 
between weight and molecular norms. 

The molecular statement is probably 
more effective for classroom purposes, and 
is clearly more expeditious for the purpose 
originally proposed by Niggli, i.e., to 
permit ready transformation from kata- 
to meso- and epi-norms. It seems quite 
unlikely, however, that in the immediate 
future extensive hand computations of 
this sort will survive as anything but 
student exercises, and in even the slowest 
electronic computers the distinction be- 
tween molecular- and weight-norm com- 
putations is trivial. The strongest argu- 
ment in favor of the molecular or cation 



per cent norm would appear to be the 
desirability of uniformity with miner- 
alogical practice; certainly in crystal- 
chemical speculations the numbers of ions 
of different species take precedence over 
their relative weights, and if a similar 
method of statement could be imposed on 
petrography without loss of information 
there would be little basis for strenuous 

In principle, it is true, there would be 
no loss. For any data manipulation based 
on weight percentages there is an analo- 
gous manipulation based on molecular or 
cation percentages. In practice, unfortu- 
nately, there would be a very considerable 
loss, for much of chemical petrology 
today operates against a background of 
phase-equilibrium studies whose results 
are reported almost entirely in graphical 
forms, the graphs being based almost 
entirely on weight per cent. Either by 
transforming field boundaries, for which 
curves usually no equations are given, or 
by converting the original data points to 
molar percentages and redrawing the 
curves, molecular versions of these dia- 
grams could be obtained. Some such 
overall transformation would be desirable, 
and for systems involving components of 
markedly different combining weights it 
would be indispensable. Whether the very 
considerable time required to effect it 
might be better used in the collection of 
new data is an open question. 

An analogous situation arises in con- 
nection with petrographic applications of 
the norm. In most reference tables, norms 
constructed with CIPW combining con- 
ventions are listed in weight per cent. In 
the most widely used compendium of this 
type, Washington's Tables, the location 
of an analysis is based on an intricate 
hierarchy of ratios computed largely from 
its norm. Unless such tables are revised 
or superseded the use of molecular norms 
in the journal literature adds a bother- 
some component of inefficiency or uncer- 
tainty to one of the routine tasks of the 
petrographer, for, although the amounts 
of individual normative minerals may not 

differ much as between weight and 
molecular norms, the differences between 
ratios computed from them may be far 
from negligible. In the Adirondack granite 
cited above, for instance, the ratio 
Q/ab is 0.785 in the weight norm and 
0.685 in the molecular norm. A sub- 
division placed at a ratio of 3:4 would 
thus classify weight and molecular norms 
computed from the same analysis in 
different groups. 

On Distinguishing Basalt from Andesite 
F. Chayes and D. Mttais 

The basis of classification in our card 
file of analyses of Cenozoic volcanic rocks 
is geographic, and until recently no pro- 
vision was made for inclusion of rock 
names on the individual analysis cards. 
When it became necessary to sort out 
particular rock types for comparison we 
were accordingly obliged to set up 
"machine definitions," definitions that 
could be utilized by a program receiving 
only the chemical analyses. For the pur- 
pose of the work reviewed in the opening 
section of this report, for instance, an 
analysis in which Si0 2 was less than 54 
per cent and the norm of which yielded a 
Thornton-Tuttle index of less than 50 
was considered a "machine basalt." The 
Thornton-Tuttle condition operates pri- 
marily to eliminate occasional highly 
feldspathoidal phonolites of the oceanic 
suite. Since in many andesites this index 
is considerably less than 50, the sole 
criterion distinguishing basalts from an- 
desites was thus silica content. A silica 
content of 54 per cent seems rather high 
for basalt and not particularly low for 
andesite; it was adopted because the 
rejection of basalts seemed more unde- 
sirable than the inclusion of andesites. 

At the time the computations were 
made, however, we had no firm estimate 
of the number of andesites that would be 
classified as "machine basalts." If these 
were sufficiently numerous our work 
would of course be open to the charge 
that we were not comparing intra- with 
circumoceanic basalts but merely some 



oceanic basalts with an unspecified mix- 
ture of andesites and basalts of the 
circumoceanic environment. 

In all the more recent data collecting 
we have made provision for inclusion of 
the name applied to each analysis in the 
source reference, and we are now adding 
this information to the older cards that 
do not contain it. In the near future the 
discriminant function calculations based 
on machine basalts will be repeated with 
"basalts as named in source references." 
In the meantime, we present indirect 
evidence that this change in the basis for 
selection will probably have little effect 
on the results. 

The evidence is drawn from the 
invaluable tabulation of analyses of 
volcanic rocks of Japan, recently pub- 
lished by Ono (1962). Of the 1057 
analyses of Cenozoic volcanics listed by 
him, 56 are of andesites containing less 
than 54 per cent of Si0 2 . Accepting this 
as a provisional estimate, probably not 
more than 50 of the 356 circumoceanic 
machine basalts in our tally were called 
andesite in the source descriptions. Be- 
cause of the scarcity of andesites among 
oceanic lavas the chance that an oceanic 
"machine basalt" is really an andesite 
must be very much less than this. The 
complementary error, exclusion of basalts 
from the machine basalt group, should be 
considerably less frequent. Nearly 200 of 
the Ono analyses are of rocks called 
basalt, and of these only 13 contain more 
than 54 per cent of Si0 2 . Because of the 
generally lower tenor of Si02 in oceanic 
basalts the incidence of this type of error 
in the oceanic machine basalt grouping 
must be very low indeed. In sum, we are 
reasonably confident that no major error 
has resulted from the use of an arbitrary 
Si0 2 value as the sole criterion for 
distinguishing between basalt and ande- 
site, but we do consider that it will be 
useful to carry through similar calcula- 
tions with groups based on rock names 
applied in source references. 

The emergence of bookkeeping devices 
that make possible the collation and 

utilization of vast numbers of rock 
analyses raises once more the difficult and 
vexing problem of nomenclature. Our 
present position with regard to both 
andesite and basalt is that the universes 
of interest are the ones whose definitions 
we piece together by examining descrip- 
tions or analyses of rocks to which these 
names have been applied in routine 
descriptive studies by many workers over 
a considerable time. It is obvious, how- 
ever, that with certain common rock 
names — granite, for instance — such a 
procedure would be little short of disas- 
trous. It is also obvious that with the 
incorporation of mineralogical informa- 
tion in the card file, a step we do not at 
present contemplate, it would be possible 
to adopt the criteria of one of the inter- 
nally consistent but often mutually 
contradictory petrographic systems. We 
would appreciate comment from inter- 
ested readers about this matter. 

Kersantites and Vogesites; a Possible 

Example of Group Heteromorphism 

D. Mttais and F. 

Older reference works (Rosenbusch, 
1887; see also Lacroix, 1933) insist that 
amphibole or biotite must be the domi- 
nant dark mineral of a lamprophyre, but 
a tendency to relax this restriction also 
became apparent very early (Iddings, 
1909), and Knopf (1936) pointedly ex- 
tended the definition to include rocks 
completely lacking hydrous minerals. For 
Knopf a lamprophyre is simply (Knopf, 
1936, p. 1748) "a mesotype or melano- 
cratic rock carrying solely ferromagnesian 
phenocrysts in an aphanitic or micro- 
granular groundmass, and in which the 
ferromagnesian minerals in the ground- 
mass show notable idiomorphism." Fol- 
lowing Iddings and Knopf, modern text 
writers sometimes leave open the possi- 
bility that hydrous minerals may be 
lacking (for a notable exception see 
Turner and Verhoogen, 1960, p. 251). In 
the examples they describe, however, 
such minerals are nearly always con- 



spicuous. In casual references varietal 
names of the group are sometimes applied 
to rocks in which the dominant ferro- 
magnesian phenocrysts are anhydrous, 
but in serious petrographic studies this 
almost never happens. (Analyses now on 
file have been drawn from more than 200 
references in only 3 of which the name of 
some one of the varieties of lamprophyre 
was used in this fashion.) Further, what- 
ever the systematic definition of the 
lamprophyre group, the definitions of 
specific varieties seem always to require 
that either mica or amphibole be a major 

Kersantites, for instance, were defined 
by Delesse (1850) as consisting essentially 
of biotite and plagioclase, the two com- 
prising about three-quarters of the rock 
(Troger, 1935), and vogesites were de- 
fined by Rosenbusch (1887) as consisting 
essentially of orthoclase and hornblende 
with variable amounts of diopside and 

plagioclase. Little is known of the nature 
of the amphibole in vogesite — of the two 
analyses so far found, one is kaersutite 
and the other is called hornblende 
(Rosenbusch, 1887) — but a number of 
analyses indicate that the biotite of 
kersantite, as, apparently, of all the 
lamprophyres, is close to phlogopite in 
composition. Both rocks usually contain 
a little altered olivine and accessory 
amounts of quartz and calcite. Kersantite 
may also contain accessory orthoclase and 
trace amounts of pyroxene. 

In work reported last year (Year Book 
62, p. 157, table 11) it was shown that the 
average compositions of kersantite and 
vogesite are virtually identical. As might 
be expected from this, the "generalized 
distance" between the two, as noted in 
the second section of part I, is very small 
for any combination of oxides. Sample 
distributions of essential oxides in the two 
varieties are listed in table 22. It will be 

TABLE 22. Observed Distributions of Essential Oxides in 103 Analyses 
of Kersantite (K) and 30 Analyses of Vogesite (V) 


Si0 2 

A1 2 3 

Fe 2 3 





Na 2 

K 2 

Ti0 2 

Origin : 






width, %: 

























K V 

K V 














7 6 












15 4 

34 5 













18 11 

33 5 
















29 9 

13 4 
















20 3 

7 7 

















6 1 
















5 1 

1 1 















1 1 


























2 1 









































40 43 46 49 52 55 53 61 64 67 

Fig. 72. Distribution of Si0 2 in kersantites 
(solid line) and vogesites (dashed line). Data of 
table 22. 

recalled that the most effective linear 
combination for discriminant analysis of 
the lamprophyre group as a whole con- 
sists of Si0 2 , MgO, and K 2 0. The 
distributions of these three oxides in the 
two groups, taken from table 22, are 
shown in figures 72-74. 

From the table and histograms it is 
clear that for every essential oxide there 
is an almost complete overlap in compo- 
sition, the rather small vogesite group 
lying wholly within the composition range 
of the considerably larger kersantite 
group. 13 There is a strong suggestion that 
the relation between the two is a kind of 
group heteromorphism, but both the way 
in which the strength of this relationship 
would have to be appraised and the 
petrological interpretation that might be 
made of it are sufficiently unusual to 
warrant a little discussion. 

Considering the petrological interpre- 
tation first, it is to be noted that in both 
rocks hydrous minerals are abundant, 
that both varieties occur as rather narrow 
dikes cutting either plutonic, metamor- 

13 Without entering into a discussion of the 
reliability of relative numbers of analyses as an 
estimate of the relative abundance of the two 
varieties, we may point out that vogesite seems 
much less common than kersantite; this ap- 
parently holds even in the type locality of 
vogesite, the Vosges mountains of eastern 

phic, or sedimentary rocks, and that they 
may even occur in the same granite or 
schist. The heteromorphism involved is 
thus not the usual contrast between 
assemblages stable in high- and low-grade 
or anhydrous and hydrous environments. 
The controls determining whether a 
mica-plagioclase or amphibole-orthoclase 
assemblage is favored must be consider- 
ably more delicate than those governing 
more strongly contrasted heteromorphic 
assemblages. Finally, because of the mode 
of occurrence of the rocks and the rather 
poor outcrop that characterizes many 
lamprophyre areas it does not seem very 
likely, though it is certainly not impos- 
sible, that detailed field study will yield 
critical information about the relation 
between the two varieties. 

Before setting out to solve the problem, 
however, we ought to be sure that it 
exists. The way in which we have treated 
the data creates a presumption that it 
does, but as a demonstration it leaves 
much to be desired. We have shown that 
the average compositions of the two 

Fig. 73. Distribution of MgO in kersantites 
(solid line) and vogesites (dashed line). Data of 
table 22. 



varieties are practically identical, and the 
same is true of the sample variances of 
essential oxides as well. We have also 
shown that for every major oxide the 
range of composition of kersantite in- 
cludes the range of composition of voge- 
site. We have not shown, however, that 
if in a vogesite-kersantite pair the Si0 2 , 
say, is in good agreement, this will also be 
true of A1 2 3 , K 2 0, Na 2 0, etc., and in the 
absence of such parallelism, heteromor- 
phism is not demonstrated. 

Of course, a persistent search might 
uncover isolated examples. One may even 
assert that if large enough samples were 
available it would certainly do so, and 
that the failure to find them indicates 
merely that the samples are too small or 
that the search has not been sufficiently 
persistent. If, however, one has to search 
hard enough for an example one can only 
conclude that it is not an example at all, 
but merely an exception. What the situ- 
ation requires is a method by which it 
would be possible to determine whether 
in kersantites and vogesites in which the 
content of some oxide is in good agree- 
ment the other oxides usually or commonly 
exhibit similar agreement. The relation 
between vogesite and kersantite seems to 
be one of group heteromorphism, but 
pending development of adequate pro- 
cedures for describing and evaluating the 
compositional similarity of the two groups 
the reader is cautioned that this is, after 
all, not much more than a guess. 






1 K 2° 






/ I 

/ 1 


7 i 


\ \. 


V 1 1 

1 1 1 


2 4 6 8 10 12 14 

Fig. 74. Distribution of K 2 in kersantites 
(solid line) and vogesites (dashed line). Data of 
table 22. 


A wide liquid immiscibility field has 
been found to transect the entire copper- 
iron-sulfur system at elevated tempera- 
tures. The new results are now being 
integrated with all previous data into a 
detailed description of the complete 
system from liquidus temperatures down 
to 100°C. 

Continued studies of the Fe-Pb-S 
system at elevated temperatures showed 
that the interruption in the tie lines be- 

tween galena and pyrite at 717° ± 3°C is 
due to the appearance of a liquid field 
that cuts the FeS 2 -PbS join at this 
temperature. The liquid field increases 
rapidly in size with increasing tempera- 
ture and cuts the PbS-FeS tie lines at 
848° ± 3°C. 

The melting temperatures are lowered 
further when sphalerite also occurs with 
the lead and iron sulfides. Since the 
temperatures during regional and contact 



metamorphism in some areas may have 
exceeded those required to melt assem- 
blages of galena, pyrite, and sphalerite in 
the appropriate proportions, the new 
findings have important bearing on our 
understanding of the formation of many 

The thermal expansion of synthetic 
pentlandite of Fe4.5Ni4.5S8 composition 
was determined over the 25°-608°C tem- 
perature range and was found to be 
several times larger than that of any 
other sulfide investigated. Pentlandite of 
Fe4.5Ni4.5Ss composition breaks down at 
610°C in the presence of vapor. Investi- 
gation of its breakdown under high con- 
fining pressure with a new differential 
thermal analysis method showed that the 
temperature of decomposition decreases 
with increasing pressure, being as low as 
525°C at 36 kb. The results indicate that 
pentlandite cannot crystallize from a 
magma and shed entirely new light on 
the mode of occurrence of this mineral in 
ores and meteorites. 

The investigations of phase relations at 
low temperatures were extended also to 
the Fe-S system. The inversion of normal 
hexagonal 58-type pyrrhotite to the low- 
temperature form, which has a hexagonal 
superstructure, was recorded by DTA 
experiments on synthetic Fei_ x S contain- 
ing from 36.445 to 37.00 weight per cent 
sulfur. Monoclinic pyrrhotite was synthe- 
sized in aqueous experiments conducted 
at temperatures up to 130°C. Two addi- 
tional phases containing water as well as 
iron and sulfur were synthesized. The 
stability field of blaubleibender covellite 
has been determined in part by similar 
aqueous methods and in part by dry 
experimentation. This mineral in the 
presence of vapor is only stable below 
157°C. Its compositional limits of stability 
lie at 67.5 and 68.5 weight per cent copper 
both at 50° and at 135°C. The phase rela- 
tions in the Cu-Ni-S system have been 
investigated at 200° and 100°C partly by 
dry and partly by aqueous experiments. 

A cooperative program was carried out 
with the Max Planck Institute and the 

University of Heidelberg for the purpose 
of investigating mineral and compound 
compositions by X-ray fluorescence and 
electron-probe techniques. A number of 
so-called "zoned bravoites" were analyzed 
by these methods. The zones were found 
to consist of authentic bravoites alter- 
nating with pyrite or vaesite. The elec- 
tron probe, when further improvements 
are made on instrumentation and meth- 
ods, may serve as a powerful tool for 
determination of the compositions of 
sulfide phases both in synthetic and in 
natural systems. 

The investigations of mineral assem- 
blages from selected ore deposits have 
proceeded considerably. About 130sphal- 
erite-pyrrhotite and 40 pyrrhotite-pyrite 
assemblages were studied from various 
levels of the Calloway Mine, Ducktown, 
Tennessee. The temperatures of forma- 
tion of ore assemblages indicated by these 
methods exhibit systematic variations 
depending not only on depth in the mine 
and position on each mining level but 
also on the associated rock types. 

The Cu-Fe-S System 
G. Kullerud 

The phase relations in this ternary 
system have been studied in considerable 
detail at 700°C and lower temperatures. 
The melting relations in the metal-rich 
portions of the system were studied by 
Schlegel and Schiiller (1952) and by 
Greig, Jensen, and Merwin {Year Book 54, 
pp. 129-134). 

The melting relations in the sulfur-rich 
portions of the system are now being 
determined by DTA experiments. In the 
Fe-S system a field of liquid immiscibility 
exists above 1082° ± 3°C and extends 
from about 46.2 to more than 95.5 weight 
per cent sulfur (Kullerud, Year Book 60, 
pp. 174-176). In the Cu-S system two 
fields of liquid immiscibility exist, one 
above 1105°C between Cu and Cu 2 S as 
reported by Heyn and Bauer (1906) and 
one above 813°C over the wide composi- 
tion range from about 27 to more than 98 



weight per cent sulfur (Kullerud and 
Yund, 1960). 

It was of considerable interest to 
determine how far these immiscibility 
fields extend into the ternary system. 
Schlegel and Schuller (1952) showed that 
the liquid immiscibility field which in the 
Cu-S system is situated between Cu and 
Cu 2 S extends very far into the Cu-Fe-S 
system and may come within 5 per cent 
of the Fe-S boundary. 

In the present study gram batches of 
the compounds CuFe 2 S 3 , CuFeS^z, and 
Cu 5 Fe$ 4 were synthesized from the ele- 
ments at 700°C. The compounds first 
without and subsequently with increasing 
amounts of added sulfur were analyzed by 
the DTA technique described by Kullerud 
(Year Book 58, pp. 161-163). In the 
experiments containing synthetic CuFe 2 S 3 
(cubanite) and sulfur a strong thermal 
effect was recorded at 865° ± 3°C when 
the total sulfur content of the charges 
exceeded about 42 weight per cent. This 
effect was very evident even when the 
charge contained as much as 95 weight 
per cent sulfur. The experiments with 
synthetic CuFeS 2 _ x (chalcopyrite) and 
sulfur showed a strong thermal effect at 
852° ± 3°C when the total sulfur content 
of the charges was in the range between 
38 and 95 weight per cent. The experi- 
ments with Cu 5 FeS 4 (bornite) and sulfur 
similarly produced a strong thermal effect 
at 840° db 3°C for all charges containing 
more than 31 weight per cent sulfur. 
These results show that the two-liquid 
field existing along the Cu-S binary 
boundary above 813°C extends increas- 
ingly further into the ternary system at 
temperatures above 813°C, and at 1083°C 
the two-liquid immiscibility field reaches 
the Fe-S boundary. The boundaries of the 
two-liquid field at 840°, 850°, and 
865°C are shown in figure 75. The points 
marked 1, 2, and 3 refer to Cu 5 FeS 4 , 
CuFeS 2 _s, and CuFe 2 S 3 compositions, 

A vertical section through the system 
from the 50 weight per cent sulfur point 
on the Cu-S boundary to the 50 weight 

46.2 wt%S 

27.0 wt °/ 

Fig. 75. Liquid immiscibility in the sulfur- 
rich portion of the Cu-Fe-S system. The points 
1, 2, and 3 indicate bornite (Cu 5 FeS 4 ), chalcopy- 
rite (CuFeS 2 -z), and cubanite (CuFe 2 S 3 ) com- 
positions, respectively. 

per cent sulfur point on the Fe-S bound- 
ary shows, as noted in figure 76, that the 
temperature of the appearance of the 
two-liquid immiscibility field at first 
increases very slowly with increasing iron 
content but at some composition corre- 
sponding to a Cu/Fe ratio between 1 : 1 
and 1:2 it rises very rapidly, finally 
reaching 1082°C on the Fe-S boundary. 
DTA experiments on Cu 5 FeS 4 on heat- 
ing show beginning of melting at 1031°C 
and complete melting at 1059°C; on 
cooling a very distinct heat effect is 
recorded at 1058°C and a very weak one 
at 1025°C. These experiments indicate 
that material of Cu 6 FeS 4 composition 
does not melt directly to a liquid of this 
composition. The liquid that begins to 
form at 1031°C contains more sulfur than 
is indicated by the Cu 5 FeS 4 formula and 
coexists with a bornite solid solution that 
contains less sulfur than is indicated by 
Cu 5 FeS 4 composition. The liquidus for 
Cu 5 FeS 4 composition is intersected at 
1058°C; above this temperature a homo- 
geneous liquid exists. The results of DTA 
experiments on Cu 5 FeS 4 and on mixtures 
of Cu 5 FeS 4 with additional sulfur are 
shown in figure 77. It is noted that sulfur 
beyond Cu 5 FeS 4 composition goes into 



solid solution in the bornite phase. 
Maximum solid solution is at 840° ± 3°C, 
where about 2.5 weight per cent sulfur 
dissolves in the bornite high-temperature 
cubic structure, corresponding to a for- 
mula of Cu 5 FeS 4 .4o. At lower tempera- 
tures the solubility of sulfur decreases 
toward the Cu 5 FeS 4 ratio. 

DTA experiments on material of 
CuFeSo composition indicated intersec- 
tion of the solidus at 860° ± 3°C and of 
the liquidus at 895° ± 3°C. In runs 
containing 10, 20, 50, 70, and 91 weight 
per cent sulfur in addition to CuFeS 2 , the 




° c 1000 




/ ~ 






in J 



CM / 




% / 

£ 900 






Fig. 76. Temperatures of appearance of 
liquid immiscibility in a vertical section through 
the Cu-Fe-S system from the 50 weight per cent 
S point on the Fe-S boundary to the 50 weight 
per cent S point on the Cu-S boundary. 

Cu 5 FeS 4 

-Wt % S added to Cu.FeS. 

Fig. 77. Melting relations in the Cu 5 FeS 4 -S 
section of the Cu-Fe-S system. Liquid immisci- 
bility exists above 840° ± 3°C. 

two-liquid immiscibility region was inter- 
sected at 850° ± 3°C. 

In DTA experiments on CuFe 2 S 3 com- 
position the solidus was intersected at 
898°C and the liquidus was recorded at 
916°C. In runs containing 20 and 90 
weight per cent sulfur in addition to 
CuFe 2 S 3 the two-liquid immiscibility field 
was intersected at 865° ± 3°C. 

These experiments show that a wide 
field of liquid immiscibility extends across 
the sulfur-rich portion of the Cu-Fe-S 
system. A liquid immiscibility field was 
shown by Kullerud (Year Book 61, pp. 
144-150) to extend across the sulfur-rich 
portion of the Fe-Ni-S system and has 
recently been found to exist in the 
Cu-Ni-S system (Moh and Kullerud, in 
preparation). From these experiments it 
appears that increasingly larger portions 
of the Cu-Fe-Ni-S tetrahedron are occu- 
pied by immiscible liquids at tempera- 
tures above 813°C and that a large volume 
of liquid immiscibility transects the entire 
system above 1083°C. 

The Fe-Pb-S System 
P. R. Brett and G. Kullerud 

The phase relations in this system at 
700°C were presented in Year Book 62. 
Above this temperature the tie lines 
between PbS and FeS 2 are broken at 
717° ± 3°C. It has now been found that 
at this temperature a liquid phase forms. 
This liquid appears at 717° ± 3°C both 
when mixtures of only PbS + FeS 2 are 
used in the experiments and when excess 
of sulfur is added. It was found that this 
initial liquid contains about 60 weight 
per cent Pb, 14 weight per cent Fe, and 
26 weight per cent S, and its composition, 
therefore, falls on or very close to the 
PbS-FeS 2 join. The liquid field increases 
rapidly in size with increasing tempera- 
ture. From figure 78, which shows the 
phase relations in the Fe-Pb-S system at 
730°C, it is noted that the liquid field 
extends about 10 weight per cent farther 
toward the sulfur corner than it did at 
717° d= 3°C. The sulfur-rich limit of the 
liquid field at 730°C is at about 55 weight 




(FeS 2 ) 

' (Fe i-.r s > 

Fig. 78. The system Fe-Pb-S at 730°C. The 
univariant assemblages, all containing vapor in 
addition to those listed, are: 1, Pb ( L> -j- Fe + 
FeS. 2, Pb(D + PbS 4- FeS. 3, PbS + Fei_*S 
+ L. 4, FeS 2 + Fei_ x S + L. 5, FeS 4- L 4- S (L) . 
6, PbS + L + S ci ). 

per cent Pb, 9 weight per cent Fe, and 36 
weight per cent S. This liquid coexists 
with liquid sulfur as noted in figure 78. 
Thus a field containing two immiscible 
liquids exists in the ternary system above 
717°C. Depending on bulk composition, 
this central liquid can coexist with pyrite 
and/or pyrrhotite or with galena and/or 

At temperatures above 730°C the 
liquid field gradually increases in size. 
The divariant region containing central 
liquid and pyrite becomes narrower and 
disappears at 743° ± 2°C, where pyrite 
melts incongruently to pyrrhotite and 
liquid of almost pure S composition. As 
this divariant region narrows, the pyr- 
rhotite-liquid divariant region as well as 
the two-liquid region becomes wider. 

The phase relations a few degrees above 
the temperature at which pyrite breaks 
down are shown schematically in figure 
79. The PbS-Fei_ x S divariant region 
narrows with increasing temperature and 
at the same time the central liquid region 
grows larger. 

At 848° ± 3°C the liquid field inter- 
sects the PbS-FeS join, and the galena- 

pyrrhotite mineral assemblage is no 
longer stable. The composition of this 
point of intersection is 71 weight per cent 
PbS and 29 weight per cent FeS. At this 
temperature the central liquid becomes 
stable with Pb (L ), and two regions of 
liquid immiscibility now exist in the 
Fe-Pb-S system. The phase relations in 
the system at a temperature slightly 
above 848°C, where the PbS-FeS tie lines 
no longer exist, are shown schematically 
in figure 80. 

At yet higher temperature the central 
liquid continues to increase in size and 
will eventually intersect the Pb ( L)-FeS tie 
lines, causing the divariant metal-rich 
region of liquid immiscibility to grow 
rather extensively. 

The second region of liquid immisci- 
bility containing central liquid + S(l> 
increases in width with increasing tem- 
peratures. At 1083° db 3°C it reaches the 
Fe-S boundary. The phase relations in the 
PbS-S portion of the Pb-S system are 
relatively unknown. From this study of 
the Fe-Pb-S ternary relations it appears 
probable that the central liquid field will 
intersect the Pb-S boundary at elevated 


Fig. 79. The system Fe-Pb-S shown sche- 
matically at about 750°C where pyrite is no 
longer a phase. The univariant assemblages, all 
containing vapor in addition to those listed, are: 
1, Pb( L ) + Fe + FeS. 2, Pb (i) 4- FeS 4- PbS. 
3, PbS 4- F ei _ x S 4- L. 4, Fe^S 4- L 4- S (i) . 
5, PbS 4- L 4- S (i ). 



(Fe, ,S) 

Fig. 80. The system Fe-Pb-S shown sche- 
matically at about 850°C where tie lines between 
galena and pyrrhotite no longer exist. The 
univariant assemblages, all containing vapor in 
addition to those listed, are: 1, Pb ( z,) + Fe + 
FeS. 2, Pb (i) + FeS + L. 3, Pb (i) + PbS + L. 
4, Fei-sS + L + S(£). 5, PbS + L + S (L) . 

temperatures and that the two-liquid 
immiscibility field will span across the 
sulfur-rich portion of the ternary system. 

The appearance of a liquid on the 
PbS-FeS 2 join as low as 717°C and on the 
PbS-FeS join at 848°C, and melting of 
PbS-Fei_ x S-FeS 2 mixtures at tempera- 
tures between 717° and 848°C, depending 
on composition, have direct bearing on 
the interpretation of the formation of 
certain ore bodies that contain galena, 
pyrrhotite, and pyrite in the appropriate 

Mixtures of galena and pyrite may have 
crystallized directly from a homogeneous 
liquid. Numerous deposits commonly 
also contain considerable amounts of 
iron-rich sphalerite, (Zn,Fe)S. Avetisyan 
and Gnatyshenko (1956), who studied the 
melting relations for various compositions 
in the FeS-PbS-ZnS plane of the Fe-Pb- 
Zn-S system, found a "eutectic" at 820°C 
and 61.5 weight per cent PbS, 30 weight 
per cent FeS, and 8.5 weight per cent ZnS. 
This indicates that the presence of 
sphalerite lowers the melting point meas- 
urably from 848°C on the FeS-PbS join 

to 820°C on the FeS-PbS-ZnS plane. 
Similar lowering of the melting point may 
be expected when ZnS is added to mix- 
tures of FeS 2 + PbS, and melting of 
sphalerite + galena + pyrite assem- 
blages therefore may begin well below 

Under regional or contact metamorphic 
conditions at localities such as Broken 
Hill, Australia, Coeur d'Alene, Idaho, and 
Bodenmais, Bavaria, the temperatures 
may have been sufficiently high to at 
least partially melt preexisting sulfides 
and to completely melt the surrounding 
sulfides in the vicinity of post-ore dikes 
such as those encountered at Broken Hill 
and Coeur d'Alene. Heating effects of this 
kind may explain intrusive relations 
between ore and dikes. 


Synthetic pentlandite of Fe4.5Ni4.5Ss 
composition as well as natural pentlandite 
from the Frood Mine, Sudbury, Canada, 
were shown by Kullerud (1963) by DTA, 
high-temperature X-ray powder diffrac- 
tion, and quenching experiments in rigid 
silica tubes to be stable only below 610°C. 
At this temperature pentlandite under its 
own vapor pressure breaks down to 
(Fe,Ni)i_ s S + (Ni,Fe) 3±x S 2 . The (Fe, 
Ni)]_ x S (pyrrhotite) phase contains less 
than 1.0 weight per cent nickel substi- 
tuting for iron, and its metal-to-sulfur 
ratio is about 9:10. The (Ni,Fe) 3±a; S2 
(high-temperature heazlewoodite) phase 
may contain several weight per cent iron, 
and its metal-to-sulfur ratio is probably 
slightly less than 3:2. 

Thermal Expansion 

N. Morimoto and G. Kullerud 

Kullerud (1963) showed that the pent- 
landite unit-cell edge is considerably 
larger at 600°C than at room tempera- 
ture. The thermal expansion of pentland- 
ite has now been measured accurately. 
Synthetic pentlandite of Fe4.5Ni4.5Ss com- 
position was ground finely under acetone 
and inserted in a silica tube with a wall 




200 400 

Temperature in °C 

Fig. 81. Variation in the unit-cell length of 
pentlandite of Fe4.5Ni4.5Ss composition as func- 
tion of temperature. 

thickness of 0.01 mm and an outside 
diameter of 0.2 mm. This tube was 
evacuated, sealed, and mounted in a 
vertical position in a Unicam high-tem- 
perature X-ray powder diffraction cam- 
era. At 25°C the unit-cell edge a of the 
synthetic pentlandite was determined as 
10.066 ± 0.002 A, o which is identical to 
the value of 10.07 A given by Berry and 

Thompson (1962) for Sudbury pent- 
landite. At elevated temperatures the 
unit-cell length increases markedly. At 
220°C the a value was found to be 
10.166; at 310°C, a = 10.210; at 375°C, 
a = 10.240; at 575°C, a = 10.310; and at 
608°C, a = 10.335 ± 0.003 A. 

In figure 81 the unit-cell length a is 
shown as a function of temperature. The 
thermal increase in the a dimension of 
pentlandite is considerably larger than 
the values reported for other sulfides such 
as PbS and ZnS. Skinner (1962) reported 
an increase of 0.0764 A in the unit-cell 
length of PbS over the 25°-607°C tem- 
perature interval and an increase of 
0.0259 A in the unit-cell length of cubic 
ZnS over the 25°-602.3°C temperature 
interval. The increase in the unit-cell 
length of pentlandite over the o 25°-608°C 
temperature range is 0.269 A, which is 
3.5 times as large as that of PbS and 
about 10 times as large as that of ZnS 
over the same temperature intervals. 

When the increase in the pentlandite 
unit-cell size is expressed in volume 
versus temperature a nearly straight line 
results as shown in figure 82. The increase 
in molar volume, calculated using 6.02472 
X 10 23 mole _1 for Avogadro's number 
and Z = 4, is from 153.62 cc/mole at 
25°C to 166.27 cc/mole at 608°C. 

100 300 400 

Temperature in °C 

Fig. 82. Variation in the unit-cell volume of pentlandite of Fe4.5Ni4.5Ss composition as function 
of temperature. 



Pressure Effect on Breakdown 
P. M. Bell, J. L. England, and G. Kullerud 
Pentlandite has in the past been 
considered a mineral of magmatic origin 
together with other sulfides such as 
pyrrhotite and chalcopyrite in ores of the 
Sudbury type. The findings by Kullerud 
(1963), that pentlandite in the presence 
of vapor is stable only below 610°C, 
indicate that pentlandite may not be of 
magmatic origin unless the temperature 
of its breakdown is increased significantly 
by high confining pressures. To test the 
effect of pressure on the thermal stability 
of pentlandite, synthetic material of 
Fe4.5Ni4.5Ss composition was inserted in 
gold capsules and investigated by the 
high-pressure DTA method recently de- 
veloped by Bell and England, which is 
described in a separate section of this 
report. These DTA experiments were 
conducted over a pressure range from 400 
to 36,000 bars. The results of these 
investigations are shown in the tempera- 
ture versus pressure plot of figure 83. It 
is noted that the temperature of pent- 
landite breakdown to (Fe,Ni)i_a;S and 
(Ni,Fe) 3 ±xS 2 decreases with increasing 

pressure. The curve marked I indicates 
the pressure-temperature conditions for 
the reaction 


(Fe,Ni)!_ x S + 

(Ni,Fe) 3±x S 2 

The (Ni,Fe) 3 ±xS 2 phase has the high- 
temperature (a) structure described by 
Kullerud and Yund (1962). In the pure 
Ni-S system the high-low (or a-/3) inver- 
sion in Ni 3 S 2 takes place at 556°C in the 
presence of vapor; however, when Fe 
replaces some of the Ni this inversion may 
be lowered significantly (Kullerud, 1963). 
The curve marked II in figure 83 shows 
the influence of pressure on the high-low 
(a-j8) inversion of the (Ni,Fe) 3±x S 2 phase 
in the presence of excess pyrrhotite. Using 
Fe4.5Ni4.5Ss composition in the experi- 
ments this curve cannot be determined in 
the P-T region where pentlandite is stable 
because here all (Ni,Fe) 3±x S 2 and (Fe, 
Ni)i_a;S react to form pentlandite. The 
extension of this curve toward lower 
pressures is being determined using 
(Ni,Fe) 3±x S 2 bulk compositions. Extrap- 
olation of existing data indicates that 
the high-low (a-$) inversion of the 




i£ 500 




Uncertainty in 
Temperature - Pressure 

Points of Equilibrium Reversal 
Determined by D.T.A. (Cr/AI) 

10 12 


16 18 20 22 

Pressure , kb 

26 28 30 32 34 36 38 

Fig. 83. Influence of confining pressure on the breakdown of pentlandite of Fe4.5Ni4.5S8 composi- 
tion is demonstrated by curves I and III. Curve II shows pressure effect on the temperature at which 
the a-/3 inversion in (Ni,Fe) 3±;E S2, heazlewoodite, takes place. 



(Ni,Fe) 3 ±xS 2 phase when saturated with 
Fe may take place as low as 500°C. The 
curve marked III in figure 83 shows the 
P-T relations of pentlandite breakdown 
to pyrrhotite and the low-temperature $ 
form of (Ni,Fe) 3 ± a; S2, which is analogous 
to the mineral heazlewoodite. 

Because of the large AH of the (Ni, 
Fe) 3 ±zS 2 inversion, the slope of curve I 
differs from that of curve III. The inter- 
section of curves I, II, and III is situated 
at about 535°C and 14 kb. 

This study indicates that pentlandite, 
which almost invariably is found together 
with pyrrhotite and chalcopyrite in 
nature, must have formed at tempera- 
tures below 610°C, disregarding the 
influence of copper on its stability. Be- 
cause of the extremely rapid reaction 
between (Fe,Ni)i_ x S and (Ni,Fe) 3±a; S2 to 
form pentlandite on cooling to 610°C, it 
is very difficult, even in the synthetic 
Fe-Ni-S system, to distinguish pent- 
landite formed through reaction on 
chilling from high temperature from pent- 
landite formed at constant temperatures 
below 610°C. In ores which cool over long 
periods of time, in comparison with the 
few seconds needed to chill a laboratory 
specimen, it is not possible to tell from 
polished section studies how pentlandite 

Phase Relations at 
Low Temperatures 

The Fe-S System 
G. H. Moh and G. Kullerud 

Efforts to clarify the phase relations in 
the dry systems at low temperatures are 
hampered by generally sluggish reaction 
rates. It was found, however, that 
hexagonal pyrrhotite (Fei_ x S) synthesized 
at elevated temperatures in the pure Fe-S 
system goes through a rapid inversion at 
temperatures in the 105°-140°C range. 
The hexagonal pyrrhotites were synthe- 
sized in silica tubes at 550°C from iron 
and sulfur. To assure homogeneous 
products the silica tubes were opened 
after 2 days of heating when all sulfur had 
reacted; the materials were ground finely 

under acetone and reheated at 550°C for 
5 to 6 more days. Differential thermal 
analyses performed on these pyrrhotites 
showed that stoichiometric FeS (36.445 
weight per cent S) goes through a rapid 
nonquenchable inversion at 139° ± 2°C. 
A sharp peak was recorded on the DTA 
charts at 138°C on heating and at 140°C 
on cooling. Pyrrhotite containing 36.72 
weight per cent S gives peaks at 124°C 
on heating and at 125°C on cooling. 
Pyrrhotite containing 36.92 weight per 
cent S only shows heat effect on heating. 
This effect is recorded at 104°C when the 
specimen is heated from room tempera- 
ture but appears at 110° to 112°C when 
the specimen is not cooled to room 
temperature before reheating. 

Specimens containing 37.35 weight per 
cent S give only one peak, which appears 
at 105°C on the DTA chart. According to 
the earlier literature troilite (FeS) on 
cooling inverts from a high-temperature 
normal hexagonal 58-type structure to a 
hexagonal superstructure that has a unit- 
cell volume six times as large as the 
normal cell. The a of the supercell is 
approximately equal to the long diagonal 
of the normal hexagonal cell, and the c 
of the supercell is about twice as large as 
the c of the normal hexagonal cell. 

The present study shows not only that 
the temperature of this inversion depends 
on composition but also that the rate of 
inversion is sufficiently rapid so that it 
can be detected by DTA experiments. 
Ores cool slowly, and the normal hexago- 
nal structure because of the rapid inver- 
sion cannot be preserved in natural 
pyrrhotites which formed at temperatures 
where the normal hexagonal structure 
was stable and which contain between 
36.445 (FeS) and about 37.00 (Fe .97 7 S) 
weight per cent S. 

Efforts were made to increase the 
reaction rates for formation of the stable 
phases in the Fe-S system at low tempera- 
tures. First, pure distilled water was 
added to mixtures of Fe and S in silica 
tubes and heated at 100°C for various 
periods of time. Reactions between sulfur 
and iron were not observed, but the 



surface of the iron grains was covered by 
films of nearly black products from 
reactions between iron and water. Next, 
about 0.25 weight per cent ammonium 
sulfide, (NH^S, was mixed with water 
in which 0.025 weight per cent NaCl was 
already dissolved. Sulfur is soluble in 
(NH^S, and small amounts of this 
chemical were added to bring sulfur into 
solution to facilitate reaction with the 
iron. In a series of experiments conducted 
at 100°C, stoichiometric FeS, synthesized 
and homogenized at 600°C and then 
quenched, was first heated with the 
solution discussed above for 13 days. The 
FeS appeared unchanged at the termina- 
tion of this experiment. Next a mixture 
of FeS + y$> was used. After 11 days 
some sulfur still remained, the liquid 
phase (pH ~ 8) was strongly yellow, and 
the iron sulfide remained powdery. In 
polished section the sulfide appeared 
weakly zoned and distinctly anisotropic. 
X-ray patterns show that it is monoclinic 
pyrrhotite. The 20, d values, and approxi- 
mate intensities obtained from X-ray 
powder diffractometer charts without 
internal standard are listed in table 23. 
Monoclinic pyrrhotite was also obtained 
at 100°C and 20 days with FeS + S and 
the solution described above. Monoclinic 
pyrrhotite has now been synthesized at 
100° and 130°C, and its natural equiva- 
lent has been found to break down at 
about 265°C. The thermal stability field 
of so-called intermediate pyrrhotite has 
not been determined. It has a composition 
between Fe .935S and Fe .9ooS and thus is 
situated between monoclinic and normal 
hexagonal pyrrhotite in the Fe-S phase 

In additional experiments containing 
iron filings or powdered slightly oxidized 
iron and sulfur in various proportions and 
heated at 100°C a number of phases 
formed, three of which seem to be new. 
Their X-ray powder diffraction patterns 
are different from any found in the 
literature. Preliminary investigations of 
these phases indicate that at least one is 
"hydrotroilite" of FeS-nH 2 composi- 
tion. Thus the addition of components 

TABLE 23. X-Ray Powder Diffraction Data 
on Synthetic Monoclinic Pyrrhotite 



















































FeKa radiation. 

* Diffuse reflection which could not be meas- 
ured exactly. 

other than Fe and S increases the reaction 
rates significantly, but apparently these 
components also enter into many of the 
reaction products. 

Blaubleibender Covellite 
G. H. Moh 

The study designed to outline the 
stability field of this mineral has shown 
that in the presence of vapor it can exist 
only below 157° ± 3°C. It forms slight 
solid solution both with covellite (CuS) 
and with digenite (Cu 9 S 5 ). As can be noted 
in the diagram of figure 84 the solid 
solution extends from about 67.5 to 68.5 
weight per cent Cu at 50° and 135°C. 
These values were established by observ- 
ing the appearance and the disappearance 
of phases as well as by chemical analyses 
on Cu and S. The diagram shows that 
blaubleibender covellite and covellite 
below 157°C can coexist stably as inde- 
pendent phases. "Blaubleibender covel- 
lite" is a misnomer; since the mineral is 
not a covellite, it should be given a 
different name. The open triangles in 
figure 84 indicate the compositions and 
temperatures at which mixtures of dige- 
nite and covellite were obtained; the 
closed triangles indicate syntheses of 
digenite and blaubleibender covellite or 
blaubleibender covellite-covellite assem- 



CuS + 

Fig. 84. Phase relations at low temperatures 
in the region between Cu 9 S 5 and CuS. Blaublei- 
bender covellite contains from 67.5 to 68.5 
weight per cent Cu and is stable below 157° ± 
3°C in the presence of vapor. 

blages. All circles, whether closed or half 
open, refer to hydrothermal experiments 
with ammonium sulfide, sulfur, and 
digenite at 135°C. The rectangles, whether 
open or half open, refer to experiments 
with carbon disulfide, sulfur, and digenite 
at 50°C. 

In ores, blaubleibender covellite occurs 
commonly as a secondary product in 
oxidation, leaching, or enrichment zones. 
Zies and Allen (1916) studied systems 
bearing on the processes taking place at 
various temperatures in such zones by 
treating copper-iron sulfides with CuS0 4 
dissolved in water. They obtained covel- 
lite after establishment of equilibrium in 
many of their experiments. Experiments 
similar to those by Zies and Allen were 
performed at 25° and 50°C. In polished 
sections made on the products formed 
before establishment of equilibrium blau- 
bleibender covellite was observed without 
exception. After attainment of equilib- 
rium, however, blaubleibender covellite 
did not exist; instead, covellite appeared. 
Blaubleibender covellite in these proc- 
esses is therefore only representative of a 

step in the reactions toward equilibrium. 
Blaubleibender covellite was also synthe- 
sized by treating bornite with aqueous 
solutions of ferric sulfate or cupric sulfate 
dissolved in water. 

The Cu-Ni-S System at 200° and 100°C 
G. H. Moh and G. Kullerud 

Silica-tube experiments with dry mix- 
tures of the elements and of various 
compounds of the system were under- 
taken at 200°C to clarify the phase 
relations at low temperatures. The tubes 
were opened at regular intervals, and the 
materials were ground to fine powder 
under acetone. These runs lasted 650 
days. Phase equilibria appear to have 
been obtained in the major portions of 
the system but not in the regions where 
the metal phases occur. The aNii_ x S 
phase, which occurred at 300°C as shown 
in last year's report, is not stable at 200°C, 
as is indicated in figure 85. In this diagram 
it is noted that the /3Ni 7 S 6 phase, which 
as yet has not been reported as a mineral, 
can coexist with chalcocite but not with 
digenite. Millerite can coexist with either 
chalcocite or digenite solid solution or 
both. This digenite at 200°C contains 
slightly more Cu than stoichiometric 
Cu 9 S 5 . The (Cu,Ni)S 2 phase (villamanin- 
ite) forms extensive solid solution in 
which the metal-to-sulfur ratio remains 
1:2. The Ni-rich end member in this 
pyrite-type solid solution has a compo- 
sition of CuNi 2 S 6 (a = 5.7245 A), and the 
Cu-rich end member has a composition of 
or close to CuNiS 4 . The (Cu,Ni)S 2 mix- 
crystals when coexisting with digenite and 
covellite at 200°C have the cell dimension 
a = 5.7059 A. 

Results of aqueous experiments at 
100°C showed that a number of tie-line 
changes take place between 200° and 
100°C. In addition blaubleibender covel- 
lite and djurleite appear. The phase 
relations at 100°C are shown in figure 86. 
Tie lines now exist between blaublei- 
bender covellite and the (Cu,Ni)S 2 
phase; and digenite, blaubleibender covel- 
lite, and villamaninite form a univariant 
assemblage. At this temperature digenite, 



djurleite, and millerite all can coexist, and It is interesting that millerite-chalco- 
/3Ni 7 S 6 , chalcocite, and djurleite also form cite and polydymite-djurleite do not form 
a univariant assemblage. stable assemblages. 

Covellite CuS 

Digenite Cug 
Chalcocite Cu 2 S 

Weight per cent 

Fig. 85. Phase relations in the Cu-Ni-S system at 200°C. The villamaninite phase is stable over 
a considerable composition range but is always stoichiometric. 




Covellite CuS 
"Blaubleibender Covellite ss.". 

Digenite Cu g S 5 
DjurleiteCu 1 gg 

NiS, Vaesite 

N '3 S A Polydymite 

NiS Millerite 
/3Ni ? S 6 

Ni,S 5 Heazlewoodite 

Weight per cent 

Fig. 86. Phase relations in the Cu-Ni-S system at 100°C. Djurleite (Cui. 96 S) and blaubleibender 
covellite now appear as stable phases. 

Studies of Ducktown, Tennessee, 
Ores and Country Rocks 

G. H. Moh and G. Kullerud with the cooperation 
of 0. Kingman 14 and R. Diffenbach u 

The Ducktown district of Tennessee 
with its several operating mines offers an 
opportunity for application to ores of 
results of synthetic laboratory systems. 
The ore minerals are mainly pyrrhotite, 
pyrite, chalcopyrite, magnetite, and 
sphalerite. Small amounts of galena, 
molybdenite, bornite, and other sulfides 
as well as secondary hematite (specu- 
larite) and gahnite appear sporadically. 
The ores occur as lenses in folded and 
strongly metamorphosed schists, quartz- 
ites, and graywackes considered of late 
Precambrian age. 

The common silicate minerals occurring 

14 Tennessee Copper Company, Ducktown, 

with the ores are quartz, actinolite, diop- 
side, plagioclase feldspar, hornblende, 
chlorite, micas (muscovite, phlogopite), 
and garnet (andradite). Calcite and 
occasionally ankerite occur as gangue 
minerals. Chlorite, biotite, and sericite, 
with occasional staurolite and, less often, 
kyanite, are common wall-rock minerals. 

Collection of specimens of ores and 
coexisting silicates and carbonates was 
undertaken during three visits to the 
area. Samples were obtained in the mines 
Calloway, Boyd, Eureka, Burra, Chero- 
kee, and Mary Polk, and from drill cores 
and surface outcrops of the ores. 

From the Calloway deposit 130 speci- 
mens containing sphalerite and pyrrhotite 
were obtained through systematic collect- 
ing on all mine levels and stopes and from 
accurately located drill cores. Forty of 
these specimens also contained pyrite. 
The unit-cell dimensions of all these 



sphalerites were determined from X-ray 
powder diffraction charts using CaF 2 as 
internal standard. The a value of sphal- 
erite can be influenced not only by iron 
occurring in solid solution replacing zinc 
in the mineral's crystal structure but also 
by manganese and cadmium. Chemical 
analyses on selected sphalerite samples 
gave 0.11 weight per cent Mn and 0.07 
weight per cent Cd. These amounts are 
too small to influence the a values of the 
sphalerites significantly. On the assump- 
tion that the Mn and Cd remain essen- 
tially constant in all sphalerite specimens 
their determined a values serve as 
measures of their iron contents. Applica- 
tion of the sphalerite-pyrrhotite solvus 
(Kullerud, 1953) to these specimens 
indicates the temperature of formation of 
the mineral pair under assumed equi- 
librium conditions. 

X-ray powder diffraction patterns 
showed that all pyrrhotites are of the 
hexagonal form. The c?i 2 values of 
pyrrhotites, which in 40 of the collected 
specimens coexist with pyrite, were 
measured, and the compositions were 
estimated from the established spacing 
versus composition curve for hexagonal 
pyrrhotites (Arnold, 1962). 

The sphalerite temperatures vary con- 
siderably, depending on their locality in 
the Calloway Mine. The ore dips steeply 
to the southeast, and its northeast- 
southwest strike coincides with the struc- 
ture of the host rocks. The highest 
temperatures were obtained in the north- 
east part of the mine. Decreasing temper- 
atures were recorded along strike toward 
the southwest. The lowest temperatures 
on each mining level were obtained in the 
southwest extremes where the vein con- 
figuration is complicated by folding and 
the presence of somewhat abnormal 
amounts of gangue minerals and wall-rock 
inclusions. The results of sphalerite- 
pyrrhotite temperature determinations on 
the 16th, 18th, and 20th mining levels — 
which are 1400, 1600, and 1800 feet, 

respectively, below the mine entrance — 
are shown in the diagrams in figure 87. 

The dio2 values measured on pyrrhotites 
coexisting with pyrite generally indicate 
lower temperatures than those recorded 
from sphalerite measurements. Coinci- 
dence in recordings was only obtained for 
low-temperature readings in the 300°- 
350°C region. Pyrrhotite compositions in 
metamorphosed ores commonly readjust 
to retrograde metamorphic conditions, as 
was demonstrated in the Bodenmais 
deposits of Bavaria (Schreyer, Kullerud, 
and Ramdohr, 1964). Such readjustments 
appear to have taken place in the Duck- 
town ores. 

Additional information on the pressure- 
temperature conditions existing during 
the geological history of this (highly) 
metamorphosed area may be obtained 
from the pyrite-pyrrhotite-magnetite 
mineral assemblage, which is stable below 
about 675°C (Kullerud, Year Book 56, 
pp. 198-200). This mineral assemblage is 
very common in the ores. The occurrence 
of kyanite in highly metamorphosed wall 
rocks indicates that the pressure must 
have exceeded 8 to 10 kb even if the 
temperature at the peak of metamorphism 
never exceeded about 400°C (Bell, 1963). 

The highest sphalerite-pyrrhotite tem- 
peratures recorded in these ores approxi- 
mate 600°C uncorrected for pressure. 
Considering the pressure effect on sphal- 
erite composition the pressure-tempera- 
ture conditions must have exceeded 14 kb 
and 800°C if the sphalerite-pyrrhotite 
assemblage existed as now found when 
the kyanite phase formed. Under such 
circumstances extensive liquefaction of 
the rocks involved must take place. 
Indications of such liquids have not been 
observed, and geological evidence points 
away from such high temperatures. It is 
therefore probable that the ores, at least 
in the form in which they are observed 
now, were emplaced at a later time than 
that at which high-pressure phases such 
as kyanite were stable. 



Calloway Mine 18 th Level 

Fig. 87. Distribution of temperatures, as ""derived from" sphalerite a values, on the 16th, 18th, 
and 20th mining levels of the Calloway Mine, Ducktown, Tennessee. The heavy lines indicate 
boundaries of ores that can be mined economically. 



X-Ray Fluorescence and Electron- 
Probe Analyses of Some 
Pyrite-Type Minerals 
G. H. Moh and J. Ottemann 15 

A number of minerals, such as pyrite 
(FeS 2 ), vaesite (NiS 2 ), cattierite (CoS 2 ), 
bravoite ([Fe,Ni]S 2 ), gersdorffite (NiAsS), 
and villamaninite ([Cu,Ni]S 2 ), have the 
pyrite Pa3 cubic-type crystal structure. 
The X-ray powder diffraction patterns of 
these minerals are similar, so that the 
phases, in spite of the differences in 
chemical composition, can be identified 
only after measurements of peak posi- 
tions. The minerals can be readily 
identified by X-ray fluorescence or elec- 
tron-probe analyses. The electron-probe 
technique was found to be particularly 
useful for the determination of compo- 
sitions of zoned minerals. 

Numerous ore specimens, each contain- 
ing one or more pyrite-type minerals, 
were obtained from worldwide localities. 
Pyrite, vaesite, cattierite, bravoite, and 
villamaninite were carefully separated, 
and the pure minerals were analyzed by 
means of X-ray fluorescence and electron- 
probe techniques. The X-ray fluorescence 
analyses were performed using a Siemens 
"Kristalloflex 4" machine with modified 
micro equipment. Both gold and chro- 
mium radiation were used. The electron- 
probe analyses were done with the 
"Cameca"-type equipment at the Max 
Planck Institute in Heidelberg. 


Zoned pyrites from numerous localities 
were investigated. It was found that 
small amounts of selenium substituting 
for sulfur in the pyrite structure have a 
marked effect on the color of the mineral 
in polished sections. Thus zoning effects 
are produced when Se-containing pyrite 
alternates with pure pyrite. Such zoned 
pyrites were obtained from drill cores of 
a coal seam at Dobrilugk, near Berlin. 
Many pyrites were found to be coated 

15 University of Heidelberg. 

with secondary bravoite, which also was 
responsible for zoning. 


Several specimens from the Shinko- 
lobwe deposits, Katanga, Belgian Congo, 
were analyzed. All contain cobalt in 
addition to nickel, and the Ni:Co ratio 
varies from 4:1 to 20:1. The copper and 
iron contents combined vary from 1 to 5 
per cent. The Ni:Cu ratio in some sec- 
tions was as high as 12:1 but is usually 
much lower. When the copper content is 
high the iron content is as a rule very 
low, but the selenium content then is high 
(selenio vaesite). The common mineral 
assemblages observed in polished sections 
are vaesite, pyrite, linnaeite, chalcopyrite, 
chalcocite, uraninite, and secondary co- 
vellite and uranium minerals. Gold and 
klockmannite occur in trace amounts. 


A sample from Shinkolobwe, Katanga, 
was found to contain Co:Ni:Fe in the 
20:2: 1 atomic ratio. The analysis showed 
some copper, the Co:Cu atomic ratio 
being about 30:1. 


Primary bravoites deposited from low- 
temperature hydrothermal solutions were 
investigated. One sample from Mina- 
ragra, Cerro de Pasco, Peru, occurs with 
vanadium ore minerals, most commonly 
patronite, VS 4 . This bravoite contains 
Ni : Fe in the 1 : 1 atomic ratio. A specimen 
from Marl Huls coal mine near Reckling- 
hausen, Ruhr, Germany, contains pyrite 
and bravoite zones. The pyrite, which is 
the dominating mineral, has very little, 
if any, nickel. The bravoite has Fe : Ni in 
about the 1 : 1 atomic ratio. A concentrate 
was made from a specimen containing 
bravoite from Mechernich, Eifel, Ger- 
many. Some of the crystals in the 
concentrate are zoned, and the composi- 
tion varies strongly from one zone to 
another. Apparently the zones represent 
mixtures of almost pure pyrite and vaesite 
with Ni:Co:Fe about 10:2:1. Copper, 



i . _ " m . _T __._ 


_ .. ! | .. _ J. _ 

: i 


ill ' ' 

! i ! ! Syntlhirt'e 

1 . i ! i . 1 1 ! 

j 1 i i i j i | 'i 

i j | • prtijlSpl 

| 1 -| ! I i i j ! 

s i i i I 1 ! 

nil i 1 ! ! i 

! ! ! ! 1 • 1. 1 

1 ! i 1 I ! 1 i 1 1 

; i 1 i i ! 1 1 1 •! ! ! 1 

! ! j ; i ; i ! i i ! ! i 

j | i ! •! i 1 : i i i ! i 

i 1 ! 1 ! i i • i i ! M 1 1 M 

! | ! j j !::!::! ! 

; 1 hiLl 1 i i l i ! 1 1 I ! 

, 1 j , ■ j | 

1 i 1 *?h i ll I . i i s i i . 

! t : ! 

! 1. 1 1 1 i-i i ! ! i 1 ! i 1 1 | 1 I 1 1 : 

a hT^ 

1 ! Wkj|j : ! 1 1 1 i.l I 1 j |j 1 I I Ij.j 

l __ VV 3 

_i c£3— £- ■ 

~- - K T 7, ml --' - •■- - 


1 w Kf m I 




I j 

Ml V llama ninite 

; i i i i 

! 1 ! : 1 1 1 

! 1 ! ! i ! 

Mils 1 i ! 

! 1 i M M i ! i j i 

M i M ! M 

•i I | | : I 1 ! 1 ■! ! M 

! ■ ! 1:1 

! 1 : i 1 : ' i 

i ' ; { i M i M | II 

1 j . i j : ! : 1 

j ! 1 1 ! 1 j ! M | | j i 1 i .1 I 

j ! ; • ; : i i ; • ! ! 

M i 1 1 ::•::!' M ' j I ^_ : : : ! 1 j 1 

! i i ! ! i 1 i : II 

^*A*P M ! 4 i ' M ! 

>-4 l - ■ |^k4 

jl / 7 : j f M | ; ! 1 • j 

fAu, Jl \ r ft ^W "IT "7 ~*F*ncc- - 


IP'l|[WPIlWlWPi*||' , IU '*■ WPJTI HJPffl W WIPPWIfflW 1 





J ' 










' 1 



co k 







! j 



i j 

! S *i 

! i 

j : 

f f 






1 | ; 

j i 


i ! 1 



J i 

i II i 

! ! ! i 



I j 


i i 

j ! 

i i 1 i ! 1 

i . 

! M j 

! ij! 

! i 1 


I I 1 i | 1 i 

M ! i 1 | 

' ' c 

; ! 1 ! 

1 1 j 

i is 


j | i 1 j 


1 Ti"=!" I"' 

: ; i i i I 



1 j ; 


M ! ! ; i 


1 ! M 1 j 

if : M | 

i ._ 

-1 1 

!co K 

1 ! | 




• i i i i i ' ! ; 

i 1 1 i 

! ! ! i i 

U U-!-fj^i 


_;_:.__ . ... — ; — ._. 

r — r— p-j -j- il-i -j-f- 

■fciza&Mh i 

• i jy 

- £fcj - - 

■ i 

j l ! '':' i : T j" :""? V»l" :' "| 

!25i i i I30i i i 35! 1 1 1401 I i Itti 


I 50 

1 155 




rfcfr | 

Fig. 88. X-ray fluorescence charts, made with gold radiation, of synthetic CuNi 2 S 6 (top), villa- 
maninite (middle), and seleniovaesite (bottom). 



arsenic, and occasionally palladium occur 
in trace amounts. Specimens from Fred- 
erickstown, Missouri, have bravoite to- 
gether with linneaite, chalcopyrite, py- 
rite, marcasite, sphalerite, galena, and 
blaubleibender covellite. The bravoite 
occurs in alternating zones with vaesite, 
which contains appreciable cobalt. The 
Co:Ni atomic ratio appears to vary 
between 1 : 4 and 1:3 in this vaesite, 
which also contains 2 to 5 weight per cent 


Villamaninite, (Cu,Ni)S 2 , was first 
described from Providencia Carmenes, 
Spain, by Schoeller and Powell (1920) but 
was later discredited by Thomson (1921). 
More recently Ramdohr (1937, 1960) and 
Hey (1962) have verified the validity of 
villamaninite as a mineral species. 

Several sections containing villamanin- 
ite from the original Spanish locality 
were investigated. Considerable variation 
in composition was recorded from one 
specimen to another. The Cu : (Ni + Co 
+ Fe) ratio varies from 1:2 to 1:1. 
Selenium occurs as a trace element when 
this ratio is low and is absent when the 
ratio is near 1:1. Of the Ni, Co, and Fe 
elements, Ni is most abundant. The 
Ni : Co atomic ratio varies from 4 : 1 to 
almost 2:1, and the Ni:Fe atomic ratio 
is >5:1. These ratios are in agreement 
with earlier values in the literature. 

The diagrams of figure 88 show X-ray 
fluorescence charts made with gold radi- 
ation on synthetic CuNi 2 S 6 (Moh and 
Kullerud, Year Book 62, pp. 189-192), 
selenio vaesite from Shinkolobwe, and 
villamaninite from Spain. 

In figure 89 an electron-probe Ka- 
intensity profile (Fe,Ni,Co) demon- 
strates zoning between pyrite and bravo- 
ite in a specimen from Maubach, near 
Diiren, Germany. High iron content indi- 
cates pyrite, and high nickel content 
indicates bravoite. In the middle of the 
section it is noted that Co substitutes for 
Ni in bravoite. 

In figure 90 a photomicrograph showing 

zoning between pyrite (white) and bravo- 
ite (gray) in a specimen also from 
Maubach has been magnified to fit with 
the Ka-intensity profile (Fe,Ni,Co,S) 
obtained with the electron probe. The 
probe track is seen as a gray diffuse line 
across the photomicrograph. 

In yet another sample from Maubach 
the concentration of sulfur, iron, nickel, 
and cobalt was measured independently 
at 2.5-micron intervals across a zoned 
pyrite-bravoite specimen as shown in the 
lower part of figure 91. The photomicro- 
graph in the upper part of this figure was 
made to illustrate where the measure- 
ments were performed. 


10 20 30 40 50 60 


Fig. 89. Electron-probe Ka intensity profile 
showing distribution of Fe, Ni, and Co in zones 
containing pyrite and bravoite. 

Plate 1 

Geophysical Laboratory 


20 30 40 


Fig. 90. Zoning between pyrite (white) and bravoite (gray) is demonstrated optically in the 
photomicrograph (top) and by the distribution of Fe, Ni, Co, and S in the electron-probe Ka intensity 
profile (bottom). 


Geoph ys ical Laboratory 


. Sulfur 

Ir on 

Mi cromeler 

Fig. 91. Zoning between pyrite and bravoite is shown in the photomicrograph (top). Contents, 
in these zones, of S, Fe, Ni, and Co as determined at 2.5-micron intervals by electron-probe meas- 
urements are diagrammed below. 




P. Ramdohr 

One of the principal scientific argu- 
ments advanced for exploration of the 
moon is the expectation that much will be 
learned about the origin and history of 
the solar system. Study of meteorites 
promises to provide information at least 
equally relevant and in many ways more 

Samples of extraterrestrial origin fall on 
earth at frequent intervals and provide 
samples from vast volumes of space. 
Study of these objects has been sporadic 
and uneven. At the moment there is 
increased activity, especially in study of 
effects of cosmic rays on the meteorites 
and in chemical investigations. In the 
past there have been some mineralogical 
studies particularly on the iron and 
stony-iron meteorites. There has been 
little systematic investigation of the 
opaque minerals in stony meteorites. 
Recent advances in techniques, notably 
in the preparation of polished sections, 
have now made these minerals accessible 
to optical examination. With the new 
preparations it is possible to study fine 
textures and to identify minerals present 
in only trace amounts. As a result 
numerous minerals known from earth 
have been observed for the first time in 
meteorites; in addition, about twenty 
entirely new minerals have been dis- 
covered, including bizarre sulfides and 
arsenides, for example (Mg,Mn,Fe)S. 

A study of opaque minerals in stony 
meteorites was begun at this laboratory 
in the winter of 1960-1961. Since then a 
total of 340 polished sections of 240 
different falls and finds have been investi- 
gated and described. All these observa- 
tions have now been compiled into a 
manuscript which together with nearly 
300 photomicrographs is scheduled for 
publication by the Smithsonian Institu- 
tion. All polished sections from which the 
photographs and descriptions were made 
have been placed in custody of the 

Smithsonian Institution, where they will 
be available to qualified investigators. 

During the past year most of the effort 
has been devoted to compilation of 
observational data, although some new 
observations have been added. Increased 
emphasis has been placed on investigation 
of enstatite achondrites, mesosiderites, 
and inclusions of sulfide nodules in iron 

Of the plethora of observational results 
it is possible to mention only a few in a 
short report. The observed similarity in 
the minor constituents of eucrites and 
mesosiderites establishes a clear relation- 
ship between these two types of meteor- 

It was noted that the mineralogy in 
areas surrounding many nodules in iron, 
whether they consist of troilite or other 
sulfides, is related to that observed in 
similar areas in meteorites of the types 
discussed above. In such areas daubree- 
lite, sphalerite paramorphic from high- 
temperature wurtzite, and alabandite- 
type mix-crystals occur. The relations 
between troilite and daubreelite are of 
special interest. These minerals form 
solid solution with each other at high 
temperatures, and unmixing often takes 
place when the temperature decreases. In 
some meteorites daubreelite and troilite 
occur as essentially pure phases in 
neighboring grains without sign of mix- 
crystal formation. In most meteorites, 
however, beautiful exsolution textures are 
plainly seen in polished sections. 

Occasionally independent idiomorphic 
crystals of daubreelite are observed adja- 
cent to troilite containing exsolved dau- 
breelite. Because of the common excess of 
troilite it should be possible to determine 
the temperature at which unmixing 
started as well as to estimate a minimum 
temperature for the formation of the 
troilite-daubreelite mineral assemblage, 
provided that quantitative information 



on the FeS-FeCr 2 S 4 system can be 
obtained. Vogel and Heumann (1950) 
suggested a reaction 2FeS + 2CrS — > 
FeCr 2 S 4 + Fe involving free iron. I have 
in no section observed free iron formed 
through such a reaction; I tentatively 
assume that FeS-FeCr 2 S 4 forms a 
(pseudo) binary join with complete solid 
solution. The position of the solvus is not 
known. The chromium monosulfide com- 
ponent is of the Cri_ x S type. Since iron 
occurs in excess the iron sulfide is troilite 
(FeS) and not pyrrhotite (Fei_J3). 

Reduction of ilmenite to rutile and iron 
(FeTi0 3 -> Ti0 2 + Fe + 0) is observed 
in every mesosiderite and is not rarely 
seen in polished sections of chondrites. In 
some sections a new iron titanium oxide 
was observed. It clearly formed at low 
temperatures and is distinctly different 
from any known terrestrial mineral. 

It was found that the complex sulfide 
solid solution which has NaCl structure 
and which was described in Year Book 61 
contains only minute amounts of calcium. 
The main constituents aside from sulfur 

are manganese, iron, and magnesium, and 
the phase is thus a magnesian-iron- 
alabandite. The Mn : Fe : Mg ratio varies. 
Usually magnesium predominates, but 
occasionally manganese does. Chromium 
may be present in amounts up to perhaps 
15 per cent. The exsolution textures 
imply complicated paragenetic relations. 
In recent months a considerable num- 
ber of meteorite specimens have been 
acquired, thus permitting an increasing 
number of falls and finds to be examined. 
The meteorites of which samples have 
been obtained most recently are Abee, 
Achilles, Arcadia, Athens, Barea, Broken 
Bow, Cumberland Falls, Dalgety Downs, 
Djati Pengilon, Enon, Felix, Great Bear 
Lake, Hedeskoga, Hoba-West, Idaho- 
Iron, Kapoeta, Khairpur, Kuttipura, 
Mincy, Morristown, Obernkirchen, Pat- 
war, Peace River, Persimon Creek, 
Pesjanoe, Pinnaroo, Pultusk, Rich Moun- 
tain, Rush Creek, Shallowater, Shaw, 
Shergotty, Siena, Soroti, Steinbach, Tie- 
schitz, Vaca Muerta, Weatherford, Wi- 


G. Kullerud and H. S. Yoder, Jr. 

New experiments have been performed 
on reactions between anhydrous and 
hydrous iron-containing silicates and 
sulfur. The reactions involving the anhy- 
drous silicates, predicted on the basis of 
last year's results, were confirmed experi- 
mentally. On the other hand, the reac- 
tions between hydrous silicates and sulfur 
yielded some unexpected products, oxy- 
hornblende and oxymica. These phases 
have been observed in ore deposit aure- 
oles. The properties of these silicate 
products will be useful in ore prospecting, 
and the reactions themselves may consti- 
tute a new approach to ore beneficiation. 

Anhydrous Silicates 

In last year's report a series of reactions 
were performed between certain iron- 
containing silicates and sulfides at 800°C 

and 2000 bars. Under such conditions the 
most common sulfide, pyrite, is not stable 
(Kullerud and Yoder, 1959), and so 
similar experiments have now been done 
with the same reactants and the same 
pressure but at 650°C, where pyrite is a 
stable phase. Under these conditions 
pyrrhotite (Fei_ x S) or troilite (FeS) is the 
stable sulfide in addition to pyrite (FeS 2 ) ; 
wiistite (FeO), magnetite (Fe 3 4 ), and 
hematite (Fe 2 3 ) are the stable oxides; 
and fayalite (Fe 2 Si0 4 ) is the stable sili- 
cate, for example. 

The reactions at this temperature are 
considerably slower than those reported 
at 800°C ; 2 weeks, however, was sufficient 
to establish phase equilibria in most 

Under the conditions outlined fayalite 
reacts with sulfur as noted in figure 92. 



560 < T < 675 
2 00 bars 


decreases when the S/Fe 2 Si0 4 mole ratio 
increases, and when it equals 1 the follow- 
ing reaction takes place: 

Fig. 92. Phase relations in the Fe-S-0-Si0 2 
system in the 560° to 675°C temperature range 
at 2000 bars. For simplicity in presentation the 
solid solutions among some of the phases, such 
as Fei_ z S, are not shown in this diagram. 

For simplicity in presentation of the 
diagram of the figure and of equations 
1-14 pyrrhotite is referred to as FeS and 
is considered stoichiometric although it 
forms significant solid solution with FeS 2 . 
When the mole fraction of sulfur is less 
than J£, the univariant assemblage 
fayalite + pyrrhotite + magnetite + 
quartz occurs. The pyrrhotite in this 
assemblage approaches stoichiometric 
composition. When the mole fraction of S 
in the Fe 2 Si0 4 + S starting materials 
equals Y, the following reaction takes 
place : 

2Fe 2 Si0 4 + S -» FeS -f 

Fe 3 4 + 2Si0 2 (1) 

which is identical to the reaction observed 
at 800°C described in last year's report. 
Addition of sulfur beyond the Yi sulfur- 
to-fayalite mole ratio at 650°C leads to 
formation of pyrite as noted in figure 92. 
Pyrrhotite + pyrite + magnetite + 
quartz form a univariant assemblage 
which is stable when Y^ < S/Fe 2 Si0 4 < 1. 
In this range the amount of pyrrhotite 

2Fe 2 Si0 4 + 

2S -» FeS 2 + 

Fe 3 4 + 2Si0 2 


The amount of magnetite decreases 
when sulfur is added beyond the 1 : 1 mole 
ratio, and hematite appears and increases 
in amount with increasing sulfur. The 
pyrite + magnetite + hematite + quartz 
univariant assemblage as also noted from 
figure 92 is stable when 1 < S/Fe 2 Si0 4 
mole ratio < 4/3. When this ratio is equal 
to 4/3, magnetite is no longer present and 
the following reaction takes place: 

3Fe 2 Si0 4 + 4S -> 2FeS 2 + 

2Fe 2 3 + 3Si0 2 (3) 

When sulfur is added beyond the 4/3 
mole ratio the amount of hematite de- 
creases and an S0 2 -rich gas appears in 
increasing amounts. When the S/Fe 2 Si0 4 
mole ratio equals 5, hematite can no 
longer exist as a phase and the following 
reaction takes place: 

Fe 2 Si0 4 + 5S -> 2FeS 2 + 

Si0 2 + S0 2 (gas) (4) 

The gas phase is referred to as having 
S0 2 composition. In reality its composi- 
tion is not exactly known but lies on or 
near the S-0 join and is uniquely defined 
at any given P and T. 

Experiments were also performed at 
650°C and 2000 bars on a natural olivine 
containing nearly equal amounts of iron 
and magnesium. When less than 20 mole 
per cent sulfur is present it reacts with a 
portion of the olivine (FeMgSi0 4 ) and the 
univariant assemblage forsteritic olivine 
+ pyrrhotite + magnetite + iron-rich 
enstatite is stable. When the amount 
of sulfur reaches 20 mole per cent all 
olivine reacts as indicated in the fol- 
lowing equation: 

4FeMgSi0 4 + S -> FeS + 

Fe 3 4 + 4MgSi0 3 (5) 

The addition of sulfur beyond 20 mole 
per cent leads to formation also of pyrite, 



and in the region containing from 20 to 
333^ mole per cent sulfur the univariant 
assemblage pyrrhotite + pyrite + mag- 
netite + iron-rich enstatite is stable. The 
amount of pyrrhotite in this region de- 
creases with increase in sulfur and is 
absent when S = 333^ mole per cent. The 
reaction for this composition may be 
expressed as 

4FeMgSi0 4 + 2S -> FeS 2 + 

Fe 3 4 + 4MgSi0 3 (6) 

If more than 333^ mole per cent but less 
than 40 mole per cent sulfur is reacted 
with such an olivine the univariant 
assemblage pyrite + magnetite + ensta- 
tite + hematite is stable. The amount of 
magnetite decreases with increasing sulfur 
in this range and is absent when S = 40 
mole per cent. The reaction for this 
composition is 

3FeMgSi0 4 + 2S -» FeS 2 + 

Fe 2 3 + 3MgSi0 2 (7) 

If more than 40 mole per cent sulfur 
reacts with olivine the amount of hema- 
tite decreases and S0 2 (gas) is produced. 
When the amount of sulfur reaches 71% 
mole per cent, hematite can no longer 
exist as a phase and the following reaction 
takes place: 

2FeMgSi0 4 + 5S -+ 2FeS 2 + 

2MgSi0 3 + S0 2 (gas) (8) 

The role of sulfur in certain metamor- 
phic processes was indicated by the 
results of experiments with hedenbergite 
(CaFeSi 2 6 ) and iron cordierite (Fe 2 Al 4 - 
Si 6 0i 8 ). Experiments at 650°C and 2000 
bars indicate the following reactions be- 
tween hedenbergite and sulfur: 

4CaFeSi 2 6 + S -> FeS + 

Fe 3 4 + 4CaSi0 3 16 + 4Si0 2 (9) 

4CaFeSi 2 6 + 2S -> FeS 2 + 

Fe 3 4 + 4CaSi0 3 16 + 4Si0 2 (10) 

3CaFeSi 2 6 + 2S -> FeS 2 + 

Fe 2 3 + 3CaSi0 3 + 3Si0 2 (11) 

16 This wollastonite may contain some iron in 
solid solution. 

2CaFeSi 2 6 + 5S -> 2FeS 2 + 

2CaSi0 3 + 2Si0 2 + S0 2 (gas) (12) 

Cordierite containing about equal 
amounts of Mg and Fe reacts with S at 
650°C and 2000 bars to form pyrrhotite, 
magnetite, Mg-rich cordierite, sillimanite 
or mullite (Al 2 SiO B ), and quartz when 
S < 20 mole per cent. Most of the Fe in 
the cordierite reacts when S = 20 mole 
per cent according to the reaction 

4FeMgAl 4 Si 5 18 + S -> FeS + 
Fe 3 4 + 2Mg 2 Al 4 Si 6 18 + 

4Al 2 Si0 5 + 6Si0 2 (13) 

With increasing amounts of sulfur, pyrite 
becomes stable and the amount of 
pyrrhotite decreases. When S = 33^ 
mole per cent, pyrrhotite is no longer 
stable and the reaction is 

4FeMgAl 4 Si 6 0i 8 + 2S -> FeS 2 + 
Fe 3 4 + 2Mg 2 Al 4 Si 6 0i 8 + 

4Al 2 Si0 5 + 6Si0 2 (14) 

Further addition of sulfur leads to the 
formation of hematite in addition to the 
phases resulting from (14). The amount 
of hematite increases, and the amount of 
magnetite decreases, with increasing sul- 
fur. When the amount of sulfur reaches 
40 mole per cent, magnetite is no longer 
a stable phase and the following reaction 
takes place : 

6FeMgAl 2 Si 5 0i 8 + 4S -> 2FeS 2 + 
2Fe 2 3 + 3Mg 2 Al 4 Si 5 0i 8 + 

6Al 2 Si0 5 + 9Si0 2 (15) 

Increase in sulfur beyond 40 mole per 
cent leads to formation of S0 2 (gas) and 
corresponding decrease in hematite. When 
71/7 mole per cent sulfur or more 
occurs, hematite is no longer stable. For 
71% mole per cent sulfur the following 
reaction takes place: 

2FeMgAl 4 Si 5 0i 8 + 5S -» 2FeS 2 + 
Mg 2 Al 4 Si 6 0i 8 + 2Al 2 Si0 6 + 

3Si0 2 + S0 2 (gas) 

Experiments were also conducted in 
rigid silica tubes in which fayalite, Fe-Mg 
olivine, hedenbergite, and Fe cordierite 



respectively were reacted with sulfur at 
500°C for periods varying from 1 to 10 
weeks. The intent was to compare the 
reaction rates of these phases in the 
presence of vapor with those described 
above, in which vapor is absent for most 
compositions in the system. The compo- 
sitions in these experiments, in which 
vapor is an inherent phase, were confined 
to regions where S0 2 (gas) does not form. 
SO 2 vapor at 500°C exerts a pressure too 
high to be contained by silica tubing. The 
reaction rates in the experiments were 
considerably slower than in the pressure 
runs at 650°C, in spite of the presence of 
vapor, which assures that the components 
of the reactants are free to exchange with 
one another. Although complete reaction 
was not obtained in any of the experi- 
ments even after 10 weeks, from study of 
the phases produced and estimates of 
their amounts at various periods of 
heating there is little doubt that the 
equilibrium assemblages excluding vapor 
are the same at 500°C as they are at 

Hydrous Silicates 

Many of the gangue and wall-rock 
minerals associated with ores are hydrous 
silicates which may have appeared as 
by-products of the ore-forming processes 
or may have been significantly altered 
through reactions with ore solutions. 

Micas and amphiboles, which are 
hydrous minerals, are commonly associ- 
ated with sulfide-type ores. To study the 
behavior of such minerals in the presence 
of sulfides, chemically analyzed biotite 
and ferroanthophyllite were heated with 
sulfur in rigid silica tubes at temperatures 
between 500° and 600°C for periods 
ranging from 2 days to 6 months. In these 
experiments sulfur was added to biotite 
and ferroanthophyllite in the exact 
amounts needed to convert to FeS all 
Fe ++ in the silicate structures. Unreacted 
sulfur was present at the termination of 
all the experiments. A strong smell of H 2 S 
was noticed when each tube was opened. 
In polished sections several per cent of 

pyrite and of magnetite and/or hematite 
in minute crystals was readily observed. 
X-ray powder diffraction patterns showed 
only mica and orthorhombic amphibole 
at the termination of the experiments. Of 
particular interest is the optical condition 
of some of the amphibole crystals. The 
extinction remained essentially parallel in 
vertical sections and the elongation re- 
mained positive, but the pleochroism 
varied from light brown to reddish brown 
to smoky gray. The pleochroism of the 
original crystals varied from colorless to 
light brown. The indices of refraction 
were distinctly higher after the sulfur 
treatment. The change in optical prop- 
erties suggests that the orthorhombic 
amphibole has been converted to an 
oxyhornblende. The final assemblage 
observed can be described as resulting 
from the following reactions: 

2(Mg 3 Fe 2 + 2 Al 2 )(Al 2 Si 6 )0 22 (OH) 2 + S 
-+2(Mg 3 Fe+ 2 Fe +3 Al 2 )0(Al 2 Si 6 )0 22 (OH) 

+ H 2 S 

3(Mg 3 Fe+ 2 Fe+ 3 Al 2 )0(Al 2 Si 6 )0 22 (OH)+2S 

-+3(Mg 3 Fe+ 3 Al 2 )(Al 2 Si 6 )0 22 (OH) + 

Fe 2 3 + FeS 2 

or with excess sulfur the reaction would 

6(Mg 3 Fe 2 + 2 Al 2 )Al 2 Si 6 22 (OH) 2 + 7S 

->6(Mg 3 Fe+ 3 Al 2 )(Al 2 Si 6 )0 22 (OH) + 
Fe 2 3 + 2FeS 2 + 3H 2 S 

Although the reaction has not run to 
completion, the phases resulting from the 
partial alteration of the ferroanthophyl- 
lite can be thus obtained. Similar relations 
can be deduced for the conversion of the 
biotite in the presence of excess sulfur to 
an oxymica + pyrite -f- hematite + H 2 S. 
These experiments have considerable 
application to processes observed in and 
near ore deposits. Two types of behavior 
are recognized as a result of sulfurization, 
depending on whether the minerals in- 
volved are anhydrous or hydrous. It 
appears that the anhydrous iron-bearing 
silicates, in general, are first depleted of 
iron, thereby generating more magne- 
sium- or aluminum-rich members of the 



requisite solid solution series. As a result 
of this process additional relatively iron- 
free silicate phases are evolved as well as 
iron sulfides and oxides. A second type of 
behavior resulting from sulfurization is 
the dehydrogenation of ferrous iron- 
bearing hydrous silicates in which the 
ferrous iron is converted to ferric iron and 
hydrogen is released. The ferric hydrous 
silicate low in OH is described as an 
oxyhornblende or an oxymica, for exam- 
ple. The hydrogen released combines with 
sulfur to give H 2 S in the gas phase. The 
iron concomitantly combines with oxygen 
from the silicate structure and sulfur to 
form sulfides and oxides. The kind of 
sulfide and oxide is dependent on the 
amount of sulfur as well as on the pres- 
sure and temperature. 

These observations should have con- 

siderable value in delineating ore bodies. 
The recognition of oxyhornblende or 
oxymica, for example, may be a sufficient 
indicator of the proximity of a sulfide- 
type ore body. Extensive investigations 
will be required, of course, to ascertain 
the limits of this process. Preliminary 
studies of the micas near the ore of the 
Ducktown, Tennessee, deposits and the 
amphiboles near the ore of the Sudbury, 
Ontario, mines indicate that this pro- 
specting tool has potential. 

In another regard, the results of these 
studies may be applied to the benefici- 
ation of low-grade, ferrous metal-contain- 
ing silicates. The sulfurization of such 
silicates may be a suitable process for 
converting economically important metals 
to sulfides, which then may be liberated 
by well known methods. 


Modern scintillation and proportional 
counting techniques and high-speed digi- 
tal computers are now the crystal- 
structure analyst's most powerful tools. 
Used intelligently, they can provide him 
with X-ray diffraction data far superior 
to those of less than a decade ago and 
afford him a manipulative power capable 
of examining atomic arrangements in the 
most complex crystals. 

Our principal effort is directed toward 
achieving a detailed understanding of the 
crystal structures of minerals important 
to the petrologist and of how these 
structures are modified by variations in 
chemical composition, temperature, and 
pressure. Structural information suffi- 
ciently precise to allow analysis of dis- 
ordering effects and other minor, but 
sometimes critical, structural variations 
requires the best data available; hence we 
are constantly attempting to improve our 
data collecting and computing techniques. 

Studies on synthetic mullite completed 
this year show it to be a complex defect 
structure with two oxygen sites and two 
cation sites only partially occupied. 
Analysis of the apparent atomic thermal 

vibrations has revealed an oxygen distri- 
bution that varies from unit cell to unit 
cell and has demonstrated that errors in 
atomic distributions in an assumed 
structure model are easily masked by 
unrealistic thermal parameters. This 
study has also shown that positional dis- 
order, induced in a coordination poly- 
hedron by substitutional disorder in the 
cation position, superimposes additional 
electron-density broadening onto that due 
to true thermal vibration to yield abnor- 
mally high thermal parameters — hence 
the term ay-parent thermal vibration. 

The micas are an extensive group of 
rock-forming sheet silicates, and studies 
of their structural relations are common- 
place in the recent literature. Our 
investigations of trioctahedral micas have 
led to the development of techniques 
whereby their gross structural features 
may be predicted with relative ease from 
a knowledge of composition and unit-cell 
dimensions. A three-dimensional analysis 
of the structures of coexisting muscovite 
and paragonite has demonstrated that 
the difference in Na/K ratio in the inter- 
layer positions of these micas causes only 



minor readjustments of the surface 
oxygen layers and leaves the dioctahedral 
layers essentially unchanged. This study 
also provides a structural explanation for 
the observation that solid solution of 
paragonite in muscovite is far more 
extensive than that of muscovite in 

Considerable effort is currently being 
expended on the pyroxene minerals, 
whose structural relationships, paramount 
to igneous and metamorphic petrology, 
are known in only a general way. This 
year saw completion of a structural study 
of the high-pressure mineral jadeite, 
NaAlSi 2 06, which is similar to diopside, 
CaMgSi 2 6 . Single crystals of three forms 
of ferrosilite, FeSi0 3 , synthesized by 
Lindsley, MacGregor, and B. Davis (see 
p. 174), are now being examined. Preces- 
sion photographs of clinoferrosilite show 
that its space group is P2i/c and suggest 
that its structure is analogous to that 
of clinoenstatite, MgSi0 3 . 

Further progress is reported in the 
study of the relation between crystal 
structure and crystal morphology: some 
space groups that cannot be differentiated 
by X-ray diffraction can be distinguished 
by the relative frequencies of the crystal 


A new value (0.76 A) is proposed for the 
ionic radius of lithium. It gives better 
agreement between observed and pre- 
dicted lithium-to-anion distances and 
accounts for the observed coordination 
numbers. It should replace the value of 
0.60 A found in the literature. 

For a mineral whose crystal structure 
is known and in which there is no omission 
solid solution, a new method is proposed 
for deriving the formula from its chemical 

Crystal Structure of Mullite 
Charles W. Burnham 

The detailed structural study of syn- 
thetic mullite of composition 1.92A1 2 3 - 
Si0 2 has been completed during this 
report year. Least-squares refinement of 
anisotropic temperature factors, /3 t y, for 
each atom reduced the discrepancy fac- 
tors, R, for three structural models to 
values listed in table 24. Comparison of 
these values does not afford an obvious 
choice of the "best" atomic arrangement, 
but electron-density maps combined with 
an analysis of interatomic distances and 
apparent thermal vibration ellipsoids 
strongly suggest that model 2 is superior 
to the others. 

Differences in the observed electron 
densities of sillimanite, Al 2 Si0 5 , and 
mullite (Year Book 62, p. 160) show that 
mullite has a defect structure correspond- 
ing to a hypothetical disordered silliman- 
ite in which some of the tetrahedron- 
linking oxygen atoms, O c , are missing. 
Cations whose positions are rendered 
untenable by oxygen removal take up 
positions in tetrahedrally coordinated 
sites, Al 2 *, that are unoccupied in 
sillimanite. Since the Al 2 * sites also re- 
quire coordination to O c , their occupancy 
increases O c coordination from two to 

To account for the composition of this 
mullite, 1.92Al 2 O r Si0 2 , 17 19 per cent of 
the oxygen atoms in the O c position must 
be missing, and 19 per cent of the tetra- 

17 The composition was determined by an 
electron-microprobe analysis carried out by J. V. 
Smith. Restudy of this sample with the electron 
microprobe has revised the Ti content from 
0.71 ± 0.05 weight per cent reported last year 
down to 0.47 ± 0.03 weight per cent (J. V. 
Smith, personal communication). 

TABLE 24. Mullite R Factors 

Model 1 

Model 2 

Model 3 

Unweighted R, all reflections 
Unweighted R, 567 observable reflections 
Weighted R, 567 observable reflections 
Standard error of fit, [2w(F - F c ) 2 /(m - n)] 1 ' 2 















Fig. 93. Projection on (001) of model 2 for the mullite structure. Two unit cells show the effect 
of removing one O c atom. Note that the O c atom that has become three-coordinated has shifted to 
the O c * position, not on the symmetry center at 0, 2, 0. 

hedral cations must occupy the Al 2 * sites. 
Assuming that only aluminum occupies 
these sites, the normal tetrahedral sites, 
C, will contain 50 per cent aluminum and 
31 per cent silicon. Occupation of 19 per 
cent of the Al 2 * sites requires 38 per cent 
of the oxygen atoms at O c to assume 
three-coordination, since there are two 
Al 2 * sites for each O c site (fig. 93). Only 
43 per cent of the O c sites remain two- 

In model 2, which was first suggested 
by Sadanaga, Tokonami, and Takeuchi 
(1962), the tetrahedron-linking oxygen 
atoms move from the O c sites to less 
symmetric O c * sites to become three- 
coordinated; each of the two O c * sites 
adjacent to O c becomes 19 per cent 
occupied. The most convincing proof of 
this distribution is seen in a high-resolu- 
tion electron-density plot of the region 
around O c , figure 94. The slightly concave 
contours between O c and O c * could result 
only from the presence of some oxygen 
nuclei at the O c * site. If all the remaining 
oxygens were at O c (model 3, table 25), 
the electron density would be elliptical, 

representing very large anisotropic ther- 
mal motion ; if they were all redistributed 
to O c * sites (model 1, table 25), the 
maximum electron density would not 
occur at the O c position. In these incorrect 
models, the anisotropic temperature fac- 
tors are the only variable parameters 
capable of compensating for nuclear 
distribution errors. The R values in table 
24 clearly show how well they have 
accomplished this. 

The structural picture emerging from 

TABLE 25. Atom Distribution in 1.92:1 
Mullite, Space Group Pbam 

Site Occupancy (ideal = 1.0) 



Model 1 

Model 2 

Model 3 

O ab 





O r 




O r * 














Al 2 ] 

Si ) 







Al 2 * 






y * 

Symmetry center 


Fig. 94. Electron-density distribution in the 
plane parallel to (001) at z — y 2 f° r the region 
surrounding the symmetry center at y, 0, y 2 . 
The contour interval is 2.5 e/A 3 ; the zero contour 
has been omitted. Electron-density values were 
computed at intervals of l/120th of a and b. 

detailed study of apparent anisotropic 
thermal motions in mullite is important 
in understanding silicates with defect and 
disordered structures. Comparison of the 
thermal model of O c in mullite (fig. 95) 
and sillimanite (Burnham, 1963a), where 
the atom is also two-coordinated but to 
an ordered tetrahedral cation arrange- 
ment, clearly shows the effect of cation 
disordering on the apparent thermal 
vibrations of the oxygen. In both min- 
erals the two largest principal axes lie in 
a plane approximately normal to the 
cation-oxygen-cation linkage. The rms 
displacements along these axes are 0.11 A 
and 0.13 ± 0.01 A in sillimanite and 0.14 
A and 0.15 ± 0.01 A in mullite. The third 
principal axis is directed toward the 
tetrahedral cations to within ±12° in 
sillimanite and ±32° in mullite; 18 the rms 
displacement along this axis is 0.06 ± 0.02 
A in o sillimanite and twice that, 0.12 ± 
0.01 A, in mullite. 

18 This large error is expected in mullite be- 
cause all three principal axes have approximately 
the same length. 


C'/T^ r = .084+.002A 

C (43% fMled) 

•"">( l.668±.00IA 
ySy C = 3l%Si, 50% Al, 19% 


Fig. 95. Schematic diagram showing details 
of the O c coordination. The sizes of the apparent 
thermal ellipsoids are not scaled to the inter- 
atomic distance. Principal axes, n, are indicated 
with arrows. Ellipsoid orientation errors are 
±7° for C and ±32° for O c . 

If we could examine each individual 
tetrahedron in many unit cells of mullite, 
we would find that, because of the dis- 
ordered arrangement of cations, the 
tetrahedra would not be of uniform size 
from one unit cell to the next. Those 
containing aluminum would presumably 
have a cation-O c distance close to 1.78 A, 
whereas in those containing silicon this 
distance should be close to 1.62 A (Smith 
and Bailey, 1963). There must be some 
readjustment of the oxygen or the cation 
or both from average positions as seen by 
X-ray diffraction to accommodate an 
individual aluminum or silicon. The re- 
fined positions of these atoms are thus 
averages over many unit cells of atoms 
having slightly different positions depend- 
ing on the particular cation involved. 
Since the maximum difference in oxygen 
positions will be only 0.16 A, atoms 
coordinating to aluminum and silicon will 
be unresolvable at room temperature and 
the effect will appear as a larger than 
normal apparent displacement of the 
oxygen toward (and away from) the 
cation. Hence the doubling of apparent 
thermal displacement of O c toward the 
cations in mullite as compared with 
sillimanite is, in fact, not a thermal effect 



at all, but a positional disorder effect 
induced by disorder of the cation position. 
There are corresponding increases in 
apparent displacements along cation- 
anion vectors for the other oxygen atoms 
in mullite and to a lesser extent for the 
cation as well. 

Figure 96 illustrates how apparent 
thermal displacements may account for 
positional disorder effects due to defects 
in the structure rather than to cation 
disordering. The oxygen atoms, O a & and 
Od, coordinate to the cation site, C, when 
it is filled. Because of vacancies in the O c 
position, only 81 per cent of the C posi- 
tions are occupied; the remaining 19 per 
cent of the cations are found filling the 
Al 2 * position, to which the oxygens O a & 
and Od are also coordinated. In the 
thermal model the largest principal axes 
of both atoms are essentially parallel to 
the C-A1 2 * vector. The rms displacements 
along these axes are 0.154 ± 0.002 A 
(O o6 ) and 0.135 ± 0.003 A (O d ). Such 
relatively large displacements aligned in 
this manner most probably represent an 
averaging effect due to oxygen atoms in 
slightly different positions depending on 
whether they are coordinating to a cation 

\ O c removed 

A I shifted from C site 

r 2 = . I 54± .0 02 A 

Fig. 96. Schematic diagram of the cation 
shifts that take place when O c is removed. 
Tetrahedra containing aluminum may not be 
preferentially affected (see text). All atoms are 
at z = Yi except Od, which has z = 0. Sizes of 
the apparent thermal ellipsoids are not scaled to 
the interatomic distances. Principal axes are 
indicated with arrows. Ellipsoid orientation 
errors are ±2° for O a & and O d , ±35° for O c *. 

in the C or the Al 2 * position. Since inter- 
atomic distances are computed for the 
average position, they will be slightly 
larger than the true distances, which 
could be determined only if the true 
positions of the atoms could be resolved. 
The distribution of aluminum and 
silicon between the C position and the 
Al 2 * position is subject to question. 
Sadanaga, Tokonami, and Takeuchi 
(1962) suggested that the new tetrahedral 
site, Al 2 *, preferentially contained alumi- 
num because the average Al 2 *-0 distance 
was significantly larger than the average 

TABLE 26. Mullite Interatomic Distances 

Atom Pairf 


Distance, Standard 
A Error 

Ah octahedron 

Ah "-Oat 4 1.897 0.001 

Ali-O d 2 1.940 0.001 

Oab-Oab' 2 2.458 0.003 

O a6 -Oab 2 2.890 0.001 

Oa6-O d " 4 2.704 0.002 

Oab'-Oa" 4 2.723 0.002 

(Al 2 ,Si) tetrahedron 

Cation-Oa6 1 1.711 0.002 

Cation-Od 2 1.728 0.001 

Cation-O c " 1 1.668 0.001 

Cation-O c *" 1 1.768 0.013 

Cation-O c *'" 1 1.733 0.013 

Oab-O d 2 2.759 0.002 

Od-Od 1 2.890 0.001 

O c "-O ab 1 2.784 0.002 

Oc"-O d 2 2.773 0.001 

O c *"-0« 6 1 2.534 0.013 

O c *"-O rf 2 2.976 0.011 

O c * / "-Oa6 1 3.107 0.011 

O c *'"-Od 2 2.663 0.011 

Al 2 * tetrahedron 

Al 2 *-O a6 1 1.822 0.003 

Al 2 *-O d 2 1.772 0.002 

Al 2 *-O c * 1 1.870 0.008 

Oab-Od 2 2.759 0.002 

Od-Od 1 2.890 0.001 

O c *-O a6 1 2.956 0.010 

O c *-Od 2 3.131 0.007 

f A single prime represents transformation of 
the coordinates listed in table 27 according to 
x' = —x,y' = —y,z' = z. Double primes repre- 
sent transformation according to a;" = Yi — x, 
y" — % + V, z" = 2. Triple primes represent 

transformation according to x' 

+ x, 


y, z 


TABLE 27. Mullite Atomic Parameters 







o ab 

0.3585 ± 0.0002 

0.4221 ± 0.0002 



o c 





0.4500 ± 0.0014 

0.0486 ± 0.0014 



o d 

0.1271 ± 0.0002 

0.2189 ± 0.0002 




Al 2 ,Si 

0.1488 ± 0.0001 

0.3404 ± 0.0001 



Al 2 * 

0.2616 ± 0.0004 

0.2055 ± 0.0004 



f B corresponds to an "equivalent" isotropic temperature factor, calculated from the anisotropic 
temperature factors, &-,-, according to £ e quiv. = % 2 2 &•,• (a;-a ; ) where the a; are axial vectors of 
the unit cell (Hamilton, 1959). * i 

C-0 distance. Their suggestion has been 
incorporated in this refinement, and, 
indeed, the refined cation-anion distances 
appear to bear out this distribution 
(table 26). The average Al 2 *-0 distance is 
1.809 A, whereas the average C-0 dis- 
tances range from 1.709 A for coordina- 
tion to Oc, to 1.725 A and 1.734 A for 
coordination to the O c * sites. 

Some structural features indicate, how- 
ever, that the Al 2 * site may contain some 
silicon. The equivalent isotropic tempera- 
ture factors (table 27) for both cation 
positions are equal and are furthermore 
significantly larger than those for pure 
silicon or tetrahedral aluminum in other 
refined silicate structures (Burnham, 
1964). In addition, the apparent rms 
thermal displacements of O a b, Od, and O c 
toward the Al 2 * site are 0.105 ± 0.003 A, 
0.115 ± 0.002 A, and 0.11 ± 0.02 A, 
respectively. These displacements are 
similar to that of O c toward the C position 
and suggest substitutional disorder in the 
Al 2 * site. Attempts to refine the per- 
centages of aluminum and silicon in the 
C site consistently reduced the amount of 
silicon assigned to that site, although the 
exact numbers are of little significance, 
owing to strong correlations between the 
silicon occupancy and temperature fac- 
tors in the least-squares normal equations. 
If there is some silicon in the Al 2 * site, 
the average Al 2 *-0 interatomic distance 
should be lower; this may very well be so 
if Oab and O d are posit ionally disordered. 

Finally, it would seem highly unlikely on 
statistical grounds alone that an appar- 
ently random removal of tetrahedron- 
linking oxygens, O c , could take place in 
such a way that only C positions contain- 
ing aluminum are affected. 

Composition Limits of Mullite, and 

the Sillimanite-Mullite Solid 

Solution Problem 

Charles W. Burnham 

Perhaps the most perplexing problem 
existing today in the Al 2 3 -Si0 2 system 
is determining and explaining the phase 
relationships between sillimanite and 
mullite. The composition of mullites, 
according to Agrell and Smith ^'(1960), 
ranges from 3A1 2 3 • 2Si0 2 (3 : 2) to 2A1 2 3 - 
Si0 2 (2:1). Natural mullites contain 
some Fe and Ti, and Muan (1957) has 
synthesized Fe-bearing mullites with Si0 2 
contents greater than that in 3 : 2 mullites. 
Most sillimanites are very close to Al 2 Si0 5 
(1:1) in composition; some, however, 
contain measurable amounts of Fe in 
solid solution (Skinner, Clark, and Apple- 
man, 1961). There are, at present, no 
confirmed occurrences of either sillimanite 
or mullite with compositions between the 
1:1 and 3:2 Al 2 3 :Si0 2 ratios. 

Comparison of the crystal structures of 
1.92:1 mullite (fig. 93, p. 224) and 
sillimanite (Burnham, 1963a) shows their 
remarkable similarity. The important 
differences are that the Si and Al tetra- 
hedra in sillimanite are ordered, whereas 



in mullite they are disordered. The higher 
Al:Si ratio in mullite requires an oxygen 
deficiency that leads to partial occupancy 
of additional tetrahedral cation sites 
available but unfilled in sillimanite. The 
chains of aluminum octahedra are prac- 
tically identical in the two structures, 
both as to internal geometry and as to 
relative orientations in their respective 
unit cells. 

Because of the ordered arrangement of 
tetrahedral cations, the c axis of the 
sillimanite unit cell is approximately 
double that of mullite; a plot of unit cell 
volume versus c for numerous mullites 
and sillimanite subcells (c SU bceii = c tr ue/2) 
given by Agrell and Smith (1960, fig. 4) 
clearly shows a discontinuity between the 
two minerals. 

Although the mullite crystal structure 
was determined from data for a crystal of 
1.92: 1 composition, it is only necessary to 
vary the amount of oxygen missing from 
the Or position (fig. 93), and hence the 
occupancy of the Al 2 * position, and the 
Al:Si ratio to generate structures for 
mullites of any given composition. As- 
suming random distribution of O c vacan- 
cies and Al and Si cations, the limit of 
solid solution toward silica is crystallo- 
graphically restricted to the 1 : 1 compo- 
sition, where there would be no O c 
vacancies. Such a structure corresponds 
to disordered sillimanite in space group 
Pbam with no superstructure. The exist- 
ence of this type of sillimanite has been 
suggested by Aramaki and Roy (1963) 
but has not yet been proved. 

Proceeding toward less siliceous compo- 
sitions, continual removal of O c could, 
geometrically, take place until, on the 
average, one O c per unit cell, or 50 per 
cent of the total, was removed. At this 
point all Al 2 * sites with four oxygens 
available for coordination would be 
occupied. Furthermore, all Si atoms 
would have been replaced by Al to main- 
tain charge balance, and the composition 
would be pure A1 2 3 . Thus, considering 
only crystallographic aspects, the dis- 
ordered mullite structure is theoretically 

suited to any composition between silli- 
manite and corundum, and there is no 
obvious reason why the composition 
should be limited to structures with 12.5 
per cent (3:2) to 20 per cent (2:1) of the 
Oc sites vacant. 

All mullites that have been suitably 
examined by single-crystal techniques 
show characteristic superstructure reflec- 
tions which may be sharp or diffuse. 19 
Agrell and Smith (1960), who have 
studied the distribution of these reflec- 
tions, state that neither type is restricted 
to mullites of a particular composition, 
and that their intensity depends on 
composition whereas their position, to a 
first approximation, does not. 

These reflections indicate that the true 
mullite structure is not ideally disordered. 
Either the distribution of oxygen vacan- 
cies or the Al-Si distribution or both may 
be partially ordered. If there is to be a 
structural explanation for solid solution 
between sillimanite and mullite, or the 
lack of it, it must almost certainly lie in 
the partial ordering schemes that give 
rise to the complex array of superstruc- 
ture reflections. Such schemes must also 
play a critical role in setting the rather 
unorthodox composition limits of mullite, 
if these compositions are indeed limiting. 
Since reliable intensities for the diffuse 
reflections are essentially unobtainable, a 
detailed crystallographic study of a 
mullite exhibiting sharp superstructure 
reflections is now of critical importance. 

Prediction of Mica Structures from 

Composition and Cell Dimensions 

G. Donnay, J. D. H. Donnay,™ and H. Takeda 20 

In the course of refining a mica crystal 
structure (Year Book 62, p. 165) and 
deciding what other mica structures 

19 Aramaki and Roy (1963) state that one of 
the mullites they studied showed no superstruc- 
ture reflections. Such reflections, if diffuse, are 
sometimes very difficult to detect; those ex- 
hibited by the 1.92:1 mullite appeared only on 
c-axis oscillation photographs. 

20 The Johns Hopkins University. 



should be refined, we asked ourselves the 
following question: With the knowledge 
of mica structures that we have now, how 
can we best predict the structural changes 
that should accompany compositional 
changes? The problem is to predict the 
coordinates of all the atoms for any- 
chosen composition. The calculations 
needed to be performed for a sufficiently 
large number of compositions so that 
some of the critical variables, such as K-0 
distances, could be plotted against com- 
position to indicate the trend. 

As a first step, we decided to limit 
ourselves to trioctahedral one-layer micas 
(fig. 97) and to assume the following: (1) 
regular tetrahedra, (2) octahedra flat- 
tened into trigonal antiprisms with 
constant cation-anion distance, (3) co- 
planar anions (0,OH,F) in (001) sheets, 
and (4) a linear relation between the 
average cation composition at the centers 

of the tetrahedra (or octahedra) and the 
cation- (0, OH, F) distance. Knowing the 
chemical composition, we apportion the 
appropriate metals, M and M t , to octa- 
hedral and tetrahedral positions. The 
metal-to-oxygen distances, d and d t , are 
calculated from the following distances 
(all in A units), obtained from the 
literature (International Tables for X- Ray 
Crystallography, vol. 3, 1962) : for coordi- 
nation 6, Al-0 = 1.91, Fe+ 3 -0 = 2.01, 
Mg-0 = 2.10, Fe+ 2 -0 = 2.12, Li-0 = 
2.16; for coordination 4, Si-0 = 1.62, 
Al-0 = 1.77, Fe+ 3 -0 = 1.86. The values 
for Si-0 and Al-0 come from Smith and 
Bailey (1963). Using the experimental 
cell dimension 6, together with the appro- 
priate values of d t and d , we evaluate the 
angles a and \p (fig. 98). a is the angle 
through which the tetrahedra are rotated 
about c*; \p is the inclination of the M -0 
bonds onto c* in the antiprisms. The 


K 2 (Mg,Fe,AI,Li) 6 [SiglSi, Al, Fe 

+ 3 

)J0 (0H,F) 
2 J 20 4 


( Paucilithionite), 



--I.64 (Mg 4 Li 2 

Ferriannite [Si 6 Fe 2 J 
[Si 5 AI 3 ] 

Annite CSLALl 

CSI fl l 

Fig. 97. The metal composition of the octahedral layer is shown with reference to the corners 
of an equilateral triaDgle. The subscripts for Li and Al are obtained by setting the sum of all cation 
charges equal to 44 and the sum of all octahedral cations equal to 6. The ordinate, giving d t (tetra- 
hedral metal-oxygen distance) in A units, is related to tetrahedral cation composition through the 
d t values listed in the text. Note that not all regions in this trigonal prism of mica composition are 
occupied. The reasons are chemical, lack of charge balance, and structural. 



Ferri-annite 9.40 
Ferriphlogopite 9.29 --LL$ t 

Poly lithion i te 8.97 






* : 

I M I I | I I 

60° 70 e 

Fig. 98. Nomogram giving a and \p from observed b and literature values of d t (tetrahedral metal- 
oxygen distance) and d (octahedral metal-anion distance) for any given composition. 

TABLE 28. Atomic Trimetric Coordinates in Rectangular Axes Ox, Oy, Oz' 
d t = M r O, d = M -0, c' = c sin 0, /3* = 180° - 


-W(?j-^ *«*-w(f) , -F 



z'/c' = z/c 





tan a 

(7 cos ^ + - d< 
2 c sin /3 




M ii 

i V5 

— I tan a 

4 12 

1 Vz 

(f cos \p + - d ( 

1 o 



c sin /3 


2 " 

d cos ^ + ^< 
c sin /3 



2 " 

d COS \p 

c sin (8 

y 2 

2 " 

rfo COS ^ 

c sin /3 




Note: In the oblique axes Ox, Oy, Oz, the trimetric abscissa x/a is equal to (x'/a) + (z/c)(c/a) 
cos 0* or to (x'/a) + ( l A){.z/c), according as only b/a = V3 is assumed or cos /3*_= a/3c is also 
assumed (as in our model), in which case c sin /3 may be replaced by (6 tan /3*)/3V3. 



value of \p is 54°44' in an octahedron; it 
increases with increasing flattening of the 
octahedron. Inherent in the calculations 
that led to figure 98 are the known mica 
relations b/a = V3 and (c/a) cos /3* = %, 
as well as the space group C2/m. Atomic 
coordinates and interatomic distances are 
then computed (table 23), in terms of 
d t , do, b, c' = d(001) = c sin /3, and the 
angles a and \f/, which are themselves 
functions of d t , d , and b. A sample calcu- 
lation for ferri-annite is shown (table 29), 
together with the comparison of observed 
and predicted parameters. 

The desired quantities can also be 
calculated in terms of d t , d , and b, by 
means of one additional substantiated 
assumption, = 100°. A program for the 
calculations, using this method, was 
written for the IBM 7094. The cell edge b 
was made to vary from 8.65 to 9.62 A in 
steps ranging from 0.02 to 0.05 A, 
depending on the size of papulation; d t 
varies o from 1.62 to 1.70 A in steps of 
0.02 A; and d , from 2.04 to 2.12 A in 
steps of 0.02 A (the value 2.07 A, needed 
for phlogopite, was also included). A total 
of about 900 mica structures were thus 
computed (in less than 0.01 hour). 

The variation of K-0 in terms of b is 
shown (fig. 99) for several values of d t 

(1.62 to 1.70); for each value of d t two 
limiting curves o are drawn (for d = 2.04 
and d = 2.12 A). It is interesting to note 
that, for a high value of b, say 9.4 A, high 
values of d t must be predicted and the 
K-0 distance will be large, greater than 
3.05 A, regardless of cl . On the other 
hand, a mica with b near 9.0 A cannot 

7 o 

have d t greater than 1.64 A, since the K-0 
distance would then become too small. 
For micas with a minimum d t of 1.62 A 
(silicon only filling the tetrahedra), we 
predict an upper limit of b of 9.16 A. 

The above purely geometric approach 
to detailed predictions of the crystal 
structures in solid solution series need not 
be restricted to sheet structures. Tourma- 
lines, pyroxenes, amphiboles, possibly 
also silicon-framework structures, could 
probably be studied in a similar fashion, 
once a few of their structures have been 
refined for known compositions and the 
geometric assumptions that can be made 
have been figured out. Structural min- 
eralogists who want to understand the 
reasons for observed solid solution ranges 
and changes in properties that accompany 
changes in composition may find struc- 
tural predictions helpful because they 
cannot hope to determine all the struc- 
tures in the range they wish to consider. 

TABLE 29. Predicted Ferri-Annite Structure 
Data: a = 5.43, b = 9.40, c = 10.32 A, = 100°0'; (Si,Fe+ 3 ) - O = 1.68; Fe+ 2 - (0,OH) = 2.12 A. 





10 3 A(x/a) 


10 3 A(y/b) z/c 

10 3 A (z/c) 


















7 0.171 



Si,Fe +3 










-1 0.391 









Fe +2 i 



- y 2 



Fe +2 n 




o y 2 


(Si,Fe+ 3 )- 

-O Fe +2 -0 

Fe+ 2 -OH 



i s/u 

K-0 K-0 at 







3.42 3.03 8°28' 










3.347 3.054 6°24 / 

59° 7' 

Note: A = observed — predicted. 


) I I I 1 I I I I 1 1 

— b(A)-*- 

Fig. 99. Nomogram giving the K-0 bond length from observed b and literature values of d t 
(1.62 to 1.70 A) and d (2.04 to 2.12 A) for any given composition. Examples shown by white circlets: 
A, annite (Eugster and Wones, 1962, AnFee); Fa, ferri-annite; Ph, phlogopite (lower point, Yoder 
and Eugster, 1954; upper point, Wones, 1963); Fph, ferriphlogopite (Steinfink, 1962); Po, pory- 
lithionite (Munoz and Takeda, private communication). Black circlets indicate experimental K-0 
values for ferriphlogopite (Steinfink, 1962) and ferri-annite (Donnay, Morimoto, Takeda, and 
Donnay, 1964). 

Crystal Structures of Coexisting 

Muscovite and Paragonite 

Charles W. Burnham and E. W. Radoslovich 

Sound explanations of structural con- 
trol over polymorphism and isomorphism 
in the micas are of considerable impor- 
tance to the metamorphic petrologist. 
Very few micas have been sufficiently 
studied by modern crystallographic meth- 
ods to allow detailed analysis of structural 
parameters. Since one of the better known 
mica structures is that of muscovite 
(Radoslovich, 1960), we thought that an 
analysis of paragonite, the sodium ana- 
logue of muscovite, would provide signifi- 
cant insight into the structural changes 
accompanying isomorphous replacement 
in sheet silicates. 

Last year one of us reported that 
excellent single crystals of 2Mi paragonite 
had been obtained from a kyanite schist 
from Alpe Sponda, Switzerland. Since this 
specimen also contains 2Mi muscovite, 
presumably formed in equilibrium with 
paragonite, we considered it worth while 
to carry out full three-dimensional refine- 
ments of both structures. This would 
provide the first known structural analy- 
sis of two similar coexisting minerals from 
the same hand specimen and would, we 
hoped, allow detailed evaluation of any 
variations in tetrahedral aluminum-sili- 
con distribution resulting from the change 
of K/Na ratio in the interlayer cation 
positions. Reexamination of the musco- 
vite structure assumed critical importance 
after Gatineau (1963) presented results, 



based on least-squares analysis of Rados- 
lovich's (1960) muscovite data, that 
differed from those reported by Rados- 
lovich (1960), particularly with respect to 
aluminum-silicon distribution within the 
tetrahedral layers. 

Full three-dimensional refinements 
have been carried out with 557 observable 
hkl reflections for paragonite and 619 
observable hkl reflections for muscovite; 
data for both crystals were measured 
using a single-crystal diffractometer with 
Ni-filtered CuKa radiation and a scintil- 
lation detector associated with pulse- 
height analysis circuitry adjusted to 
accept 90 per cent of the diffracted 
characteristic radiation. Unit-cell dimen- 
sions of both specimens are listed in table 
30. Least-squares refinement of aniso- 
tropic thermal models reduced the dis- 
crepancy factors, R, to 0.038 (unweighted) 
and 0.034 (weighted) for paragonite and 
0.038 (unweighted) and 0.038 (weighted) 
for muscovite. The standard error of 
fit (= [2w(F ohs - Fcai) 2 /(w - n)Y>) 
is 0.971 for paragonite and 1.305 for 
muscovite. 21 At this stage no attempt has 
been made to locate hydrogen, and the 
refinement has not been biased by any 
predetermined tetrahedral cation distri- 
bution; both crystallographically distinct 
positions have been assigned the scatter- 
ing power of fully ionized silicon. 

A partial electron-microprobe analysis 
of both mica specimens for potassium, 

TABLE 30. Unit-Cell Dimensions of Coexisting 

Muscovite (Mu 66 ) and Paragonite (Mui 5 ), 

Alpe Sponda, Switzerland* 



a, A 
6, A 

c, A 

5.174 ±0.001 

8.976 ± 0.001 

19.875 ±0.003 

95.590 ± 0.006 

5.134 ± 0.001 

8.907 ± 0.001 

19.376 ± 0.002 

94.625 ±0.006 

* Values determined by least-squares analysis 
of precision Weissenberg film measurements. 

21 The expected value of this quantity is 1.0 
for a converged least-squares analysis carried 
out using proper absolutely scaled weights for 
the observations. 

calcium, and aluminum was kindly 
undertaken by J. V. Smith. His prelimi- 
nary results show the paragonite to con- 
tain 1.80 to 1.85 weight per cent K 2 and 
the muscovite to contain approximately 
7.8 weight per cent K 2 0, with no appreci- 
able calcium present (J. V. Smith, 
personal communication). Assuming ideal 
Al/Si ratios, these results correspond to 
the formulas 

Paragonite (Ko.i5Nao.86)Al 2 (Si 3 Al)Oio(OH) 2 
Muscovite (K .65Nao.35)Al 2 (Si 3 Al)Oio(OH) 2 

Our refinements were carried through to 
the final stages assuming incorrect com- 
positions corresponding to Mun for 
paragonite and Mu 74 for muscovite. 
During the final stage of each refinement, 
the occupancy of the interlayer alkali 
positions was allowed to vary, subject to 
the restriction that the total occupancy 
of the positions is 100 per cent. The 
refinements converged to occupancies 
corresponding to Ko.15Nao.85 ± 0.02 for 
paragonite and Ko.eeNao.34 ± 0.02 for 

Comparison of the final atomic coordi- 
nates shows that the two crystallograph- 
ically independent tetrahedral cations are 
coplanar in both structures. The two 
apical oxygen atoms, O a and O&, are also 
coplanar in both structures. Two of the 
three oxygen atoms (O c , O e ) making up 
the basal triad of each tetrahedron are 
coplanar, whereas the differing z coordi- 
nate of the third oxygen (O d ) causes the 
basal plane of each tetrahedron to be 
tilted slightly. The equivalent isotropic 
temperature factors, B, are remarkably 
similar atom for atom in the two struc- 
tures, and the equality of temperature 
factors for all four tetrahedral cations 
(0.65, 0.65, 0.62, 0.63) immediately sug- 
gests that the aluminum-silicon distribu- 
tion is identical in all four positions. 

Important interatomic distances are 
listed in table 31. The 2VO and T 2 -0 
distances demonstrate conclusively that 
the distribution of tetrahedral cations is 
disordered and the same in both tetra- 
hedra in both structures. In muscovite 
the two crystallographically distinct tet- 



TABLE 31. Interatomic Distances (A) in 2Mi 
Muscovite (Mu 66 ) and Paragonite (Mui 8 ) 

Atom Pair 



T\ tetrahedron 

Ti-Oa (apical) 

1.642 ±0.004 

1.648 ± 0.002 


1.645 ±0.004 

1.655 ± 0.004 

TvO d 

1.643 ± 0.004 

1.642 ±0.004 


1.649 ±0.004 

1.664 ± 0.003 

Mean 7VO 



O a -O c 

2.694 ± 0.005 

2.706 ± 0.004 

O a -O d 

2.725 ± 0.005 

2.720 ± 0.004 

O a -Oe 

2.701 ± 0.005 

2.709 ± 0.004 


2.696 ± 0.005 

2.707 ± 0.005 

O c -O e 

2.654 ± 0.005 

2.685 ± 0.005 

o d -o e 

2.639 ± 0.005 

2.656 ± 0.005 

Mean 0-0 



T 2 tetrahedron 

Ti-Ob (apical) 

1.644 ±0.004 

1.652 ± 0.003 


1.648 ±0.004 

1.656 ±0.004 

T 2 -O d 

1.644 ± 0.004 

1.653 ±0.003 

T 2 -O e 

1.645 ±0.004 

1.644 ±0.004 

Mean T 2 -0 



O fa -O c 

2.702 ± 0.005 

2.709 ± 0.005 

b -0d 

2.726 ± 0.005 

2.726 ± 0.005 

o 6 -o e 

2.699 ± 0.005 

2.707 ± 0.005 

O c -O d 

2.647 ± 0.005 

2.677 ± 0.005 

O c -O e 

2.647 ± 0.005 

2.650 ± 0.005 

o d -o e 

2.695 ± 0.005 

2.709 ± 0.005 

Mean 0-0 



Al octahedron 

Al-0 a 

1.943 ±0.004 

1.933 ±0.002 

Al-0 a' 

1.920 ± 0.004 

1.914 ±0.002 


1.917 ±0.004 

1.906 ±0.004 

Al-0 b ' 

1.946 ±0.004 

1.938 ±0.004 


1.907 ±0.004 

1.891 ±0.004 


1.907 ±0.004 

1.899 ±0.004 

Mean Al-0 



Mean of 9 

unshared 0-0 2.824 


Mean of 3 

shared 0-0 



Interlayer cation 

K,Na-O c 

2.762 ± 0.004 

2.531 ±0.004 

K,Na-O d 

2.823 ± 0.004 

2.726 ±0. 004 

K,Na-O e 

2.795 ± 0.004 

2.668 ±0.004 

Mean K,Na-0 



rahedra are identical within the precision 
of the determination. The two tetrahedra 
in paragonite, although having identical 
average interatomic distances, are indi- 
vidually somewhat distorted as evidenced 
by comparing Ti-O d , Ti-O e , T 2 -O d , and 
T 2 -O e distances. The average T-0 dis- 
tances are less than the value of approxi- 
mately 1.655 A expected for tetrahedra 
containing 75 per cent Si and 25 per cent 
Al (Smith and Bailey, 1963). 

Comparison of interatomic distances in 
the aluminum octahedra shows that this 
layer is practically unaffected by the 
change of K/Na ratio in the interlayer 
cation position. The OH-OH shared 
octahedral edges are significantly shorter 
in both structures than the shared O a -O a 
and 0&-0& edges: 2.370 A versus 2.448 A 
and 2.443 A in muscovite, and 2.362 A 

o o ; 

versus 2.450 A and 2.439 A in. paragonite. 
These aluminum octahedra show no 
unusual distortions attributable to 
"stresses" arising in the tetrahedral layers 
or due to the presence of interlayer 
alkalies. The average Al-0 distances cor- 
respond closely to those found in other 
silicates, and distortions from ideality are 
primarily due to octahedral edge sharing 
resulting in the expected shared-edge 
contraction (see, for example, Burnham, 

Because of the marked ditrigonal 
nature of the tetrahedral sheets (fig. 100), 
the effective alkali coordination is six 
rather than twelve. The average of six 
alkali-0 distances reflects the change of 
K/Na ratio. This composition change has 
little or no effect on the relative orienta- 
tions of surface oxygens (O c , 0<*, O e ) 
between layers. 

Of critical importance, then, is the 
question: What changes do take place in 
the mica framework when sodium is 
substituted for potassium? In an overall 
sense, the answer is that there are no 
changes corresponding to first-order 
effects but that there are some slight 
shifts corresponding to second-order ef- 
fects. These manifest themselves primar- 
ily as a contraction of the surface oxygen 



Fig. 100. Projection on (001) of one tetrahedral layer of muscovite (solid lines) and paragonite 
(dashed lines). The basis for superposition of layers is exact coincidence of alkali atoms. Concentric 
circles show the relative sizes of K (outer) and Na (inner) atoms. 

network and a reduction in the interlayer 
separation. Note, in table 31, that, of the 
changes in T-0 distances, the greatest is 
in some of the T-O SU rface distances, where- 
as increases in the T-Oapicai distances are 
about equivalent to the standard errors. 
Adjustment of the surface oxygen net- 
work to accommodate more Na and less 
K is seen in the significant changes in 
some Osurface-0 S urface distances; the simul- 
taneous decrease in interlayer separation 
results in minor increases in only three of 
the six distinct O ap icai-0 S urface distances. 

Muscovite-par •agonite solid solution. Sta- 
bility relations within the pseudobinary 
system muscovite-paragonite have been 
investigated experimentally by Eugster 
and Yoder {Year Book 54, p. 125) and 
Nicol and Roy (1964). Both studies have 
demonstrated that complete solid solution 
does not exist between muscovite and 
paragonite within the temperature-pres- 
sure regions investigated. A recent study 
of coexisting muscovites and paragonites 
by Zen and Albee (1964) shows that the 

solvus is extremely asymmetric and that 
the solubility of muscovite in paragonite 
is very limited. These authors suggest 
that the maximum on the solvus will lie 
at approximately 80 mole per cent 

From a structural point of view, the 
asymmetric solid solution limits are easily 
explained by considering the variation of 
average alkali-oxygen interatomic dis- 
tances with changing K/Na atomic ratio. 
In figure 101 the averages obtained in our 
structure refinements are plotted versus 
composition. The expected average dis- 
tances for pure alkalies (in octahedral 
coordination) are given by the Inter- 
national Tables for Crystallography (vol. 3, 
p. 258) as 2.83 A for K-0 and 2.44 A for 
Na-O; Radoslovich (1960) reported an 
average K-0 distance of 2.81 A for pure 
muscovite. Figure 101 shows that substi- 
tution of Na for K does not result in a 
linear variation of alkali-oxygen distance, 
but that the change as Na replaces K is 
gradual, whereas only a small amount of 






1 I 1 







MIM — ^T^ 


^-—-~~~~ — — -*"-*""* 


"-*■"*" ^--*" 








^^ — 






/ «■»•"'" 


/ ,, ■"""^ 




-V ^~ 



o ^ 


■■-Standard error 








Fig. 101. Effect of Na-K atomic substitution on average six-coordinated alkali-oxygen inter- 
atomic distances. Two intermediate points represent average distances in muscovite (Mu 6 {) and 
paragonite (Mu i5 ). The straight dashed line emphasizes the nonlinearity of the change of average 
distance with composition. 

K substituting for Na causes an appreci- 
able increase in the average distance. 
This is, of course, an expected result, since 
an Na ion is considerably smaller than a 
K ion (Wells, 1962, p. 71), and may fit 
easily into a K coordination polyhedron, 
whereas substitution of K for Na requires 
expansion of the coordination polyhedron. 
For this reason alone, solid solution 
involving these atomic species can hardly 
be ideal. 

These structural relations lead to the 
conclusion that, at room temperature and 
pressure, paragonite will accept only a 
very small amount of K substituting for 
Na, but that muscovite will allow a 
considerably larger amount of Na to 
substitute for K, in agreement with the 
results of Nicol and Roy (1964) and Zen 
and Albee (1964). The variation of 
average alkali-oxygen distances further- 
more suggests that, in view of the minor 
changes in the mica framework between 
Mu fi 6 and Mui 5 , the structural differences 
between Mu 6 e and Muioo will be negligible. 
On the other hand, the most significant 
structural changes will exist between 
Muis and Mu (pure paragonite). 

Throughout this discussion we have 
assumed that K and Na are truly dis- 
ordered in both structures. There is, 
however, the possibility that the alkalies 
might be ordered within each interlayer 
plane. If K ions were restricted to certain 
planes, Na to others, and the sequence 
were random in the c direction, there 
would be no visible superstructure. Evi- 
dence for such an arrangement might be 
found by analyzing the apparent aniso- 
tropic thermal motion of the surface 
oxygens; if each interlayer plane con- 
tained only one kind of alkali, the largest 
apparent rms displacement of the oxygens 
should be normal to the layers (along c*), 
representing this random change in 
interlayer separation from layer to layer. 
Although a room-temperature determina- 
tion is not conclusive, the rms displace- 
ments of O c , Od, and O e toward the alkali 
are larger than they are toward the 
disordered Si 3 Al positions, and none of 
the longest principal axes is directed 
parallel to c*. The large rms displace- 
ments toward the alkali tend to uphold a 
completely random arrangement of alka- 



Charles W. Burnham 

Single crystals of clinoferrosilite, 
FeSi0 3 , synthesized by Lindsley, Mac- 
Gregor, and B. Davis (see p. 174) 
at 20 kb pressure and 1150°C, have 
been examined by precession and Weis- 
senberg photographic methods. System- 
atic extinctions correspond to those of 
space group P2 1 /c. Preliminary unit-cell 
dimensions, obtained by least-squares 
analysis of precision Weissenberg film 
data including corrections for systematic 
errors due to film shrinkage, specimen 
absorption, and camera eccentricity, are 

is examined, its loss of directional charac- 
ter is most striking : a bond ceases to be a 
vector. This observation has been stated 
before, in various ways. We spoke of it 
as "punctualization of charges with 
disregard of the sign of the charges" 
(Year Book 61, p. 130), without, however, 
recognizing all the implications of such a 
state of affairs. 

Consider the diffraction aspect P*/*. 
Donnay and Harker (1937) predicted one 
and the same morphology for the three 
space groups, Pm, P2, and P2/m, which 
belong to this aspect. (The extinction 
criteria of the diffra